comparison docs/scripts/html/code/SimilaritySearchingFingerprints.html @ 0:4816e4a8ae95 draft default tip

Uploaded
author deepakjadmin
date Wed, 20 Jan 2016 09:23:18 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:4816e4a8ae95
1 <html>
2 <head>
3 <title>MayaChemTools:Code:SimilaritySearchingFingerprints.pl</title>
4 <meta http-equiv="content-type" content="text/html;charset=utf-8">
5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css">
6 </head>
7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10">
8 <br/>
9 <center>
10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a>
11 </center>
12 <br/>
13 <pre>
14 1 #!/usr/bin/perl -w
15 2 <span class="c">#</span>
16 3 <span class="c"># $RCSfile: SimilaritySearchingFingerprints.pl,v $</span>
17 4 <span class="c"># $Date: 2015/02/28 20:46:21 $</span>
18 5 <span class="c"># $Revision: 1.18 $</span>
19 6 <span class="c">#</span>
20 7 <span class="c"># Author: Manish Sud &lt;msud@san.rr.com&gt;</span>
21 8 <span class="c">#</span>
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
23 10 <span class="c">#</span>
24 11 <span class="c"># This file is part of MayaChemTools.</span>
25 12 <span class="c">#</span>
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
29 16 <span class="c"># later version.</span>
30 17 <span class="c">#</span>
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
34 21 <span class="c"># details.</span>
35 22 <span class="c">#</span>
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
37 24 <span class="c"># along with MayaChemTools; if not, see &lt;http://www.gnu.org/licenses/&gt; or</span>
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
40 27 <span class="c">#</span>
41 28
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">&quot;$FindBin::Bin/../lib&quot;</span><span class="sc">;</span>
44 31 <span class="k">use</span> <span class="w">Getopt::Long</span><span class="sc">;</span>
45 32 <span class="k">use</span> <span class="w">File::Basename</span><span class="sc">;</span>
46 33 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
47 34 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
48 35 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
49 36 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
50 37 <span class="k">use</span> <span class="w">SDFileUtil</span><span class="sc">;</span>
51 38 <span class="k">use</span> <span class="w">StatisticsUtil</span><span class="sc">;</span>
52 39 <span class="k">use</span> <span class="w">PseudoHeap</span><span class="sc">;</span>
53 40 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsFileUtil</span><span class="sc">;</span>
54 41 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsBitVector</span><span class="sc">;</span>
55 42 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsVector</span><span class="sc">;</span>
56 43
57 44 <span class="k">my</span><span class="s">(</span><span class="i">$ScriptName</span><span class="cm">,</span> <span class="i">%Options</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="cm">,</span> <span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$TotalTime</span><span class="s">)</span><span class="sc">;</span>
58 45
59 46 <span class="c"># Autoflush STDOUT</span>
60 47 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
61 48
62 49 <span class="c"># Starting message...</span>
63 50 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
64 51 <span class="k">print</span> <span class="q">&quot;\n$ScriptName: Starting...\n\n&quot;</span><span class="sc">;</span>
65 52 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
66 53
67 54 <span class="c"># Get the options and setup script...</span>
68 55 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
69 56 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> != <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
70 57 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">&quot;$FindBin::Bin/$ScriptName&quot;</span><span class="s">)</span><span class="sc">;</span>
71 58 <span class="s">}</span>
72 59
73 60 <span class="c"># Process reference and database file names...</span>
74 61 <span class="k">my</span><span class="s">(</span><span class="i">@FingerprintsFilesList</span><span class="s">)</span><span class="sc">;</span>
75 62 <span class="i">ProcessFingerprintsFileNames</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
76 63
77 64 <span class="c"># Process options...</span>
78 65 <span class="k">print</span> <span class="q">&quot;Processing options...\n&quot;</span><span class="sc">;</span>
79 66 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
80 67 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
81 68
82 69 <span class="c"># Setup information about fingerprints inut and SD/text output files...</span>
83 70 <span class="k">my</span><span class="s">(</span><span class="i">%FingerprintsFilesInfo</span><span class="cm">,</span> <span class="i">%OutputFilesInfo</span><span class="cm">,</span> <span class="i">%SimilaritySearchInfo</span><span class="s">)</span><span class="sc">;</span>
84 71 <span class="k">print</span> <span class="q">&quot;Checking and retrieving information from reference and database fingerprints files...\n&quot;</span><span class="sc">;</span>
85 72 <span class="i">RetrieveFingerprintsFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
86 73
87 74 <span class="c"># Perform similarity search...</span>
88 75 <span class="k">print</span> <span class="q">&quot;Performing similarity search...\n&quot;</span><span class="sc">;</span>
89 76 <span class="k">my</span><span class="s">(</span><span class="i">%SimilaritySearchResults</span><span class="cm">,</span> <span class="i">%DatabaseFingerprintsFileData</span><span class="s">)</span><span class="sc">;</span>
90 77 <span class="i">PerformSimilaritySearch</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
91 78
92 79 <span class="k">print</span> <span class="q">&quot;\n$ScriptName:Done...\n\n&quot;</span><span class="sc">;</span>
93 80
94 81 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
95 82 <span class="i">$TotalTime</span> = <span class="w">timediff</span> <span class="s">(</span><span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="s">)</span><span class="sc">;</span>
96 83 <span class="k">print</span> <span class="q">&quot;Total time: &quot;</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
97 84
98 85 <span class="c">###############################################################################</span>
99 86
100 87 <span class="c"># Perform similarity search using fingerprints data in reference and database text files...</span>
101 88 <span class="c">#</span>
102 <a name="PerformSimilaritySearch-"></a> 89 <span class="k">sub </span><span class="m">PerformSimilaritySearch</span> <span class="s">{</span>
103 90
104 91 <span class="k">print</span> <span class="q">&quot;\nProcessing fingerprints data for reference molecules...\n&quot;</span><span class="sc">;</span>
105 92 <span class="i">ReadReferenceFingerprintsData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
106 93
107 94 <span class="i">InitializeSimilaritySearchResults</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
108 95 <span class="i">GenerateSimilaritySearchResults</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
109 96 <span class="i">WriteSimilaritySearchResultFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
110 97 <span class="s">}</span>
111 98
112 99 <span class="c"># Find similar molecules from database molecules for individual or multiple reference molecules...</span>
113 100 <span class="c">#</span>
114 <a name="GenerateSimilaritySearchResults-"></a> 101 <span class="k">sub </span><span class="m">GenerateSimilaritySearchResults</span> <span class="s">{</span>
115 102 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseFingerprintsFileIO</span><span class="cm">,</span> <span class="i">$FingerprintsCount</span><span class="cm">,</span> <span class="i">$IgnoredFingerprintsCount</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsObject</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ReferenceFingerprintsObject</span><span class="cm">,</span> <span class="i">$ReferenceIndex</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$FusedComparisonValue</span><span class="cm">,</span> <span class="i">@ComparisonValues</span><span class="s">)</span><span class="sc">;</span>
116 103
117 104 <span class="k">print</span> <span class="q">&quot;Processing fingerprints data for database molecules...\n&quot;</span><span class="sc">;</span>
118 105
119 106 <span class="s">(</span><span class="i">$FingerprintsCount</span><span class="cm">,</span> <span class="i">$IgnoredFingerprintsCount</span><span class="s">)</span> = <span class="s">(</span><span class="n">0</span><span class="s">)</span> x <span class="n">3</span><span class="sc">;</span>
120 107
121 108 <span class="i">$DatabaseFingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsFileIOParameters</span>}}<span class="s">)</span><span class="sc">;</span>
122 109 <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
123 110
124 111 <span class="i">@ComparisonValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
125 112
126 113 <span class="j">DATABASEFP:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;Read</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
127 114 <span class="i">$FingerprintsCount</span>++<span class="sc">;</span>
128 115
129 116 <span class="k">if</span> <span class="s">(</span>!<span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;IsFingerprintsDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
130 117 <span class="i">$IgnoredFingerprintsCount</span>++<span class="sc">;</span>
131 118 <span class="k">next</span> <span class="j">DATABASEFP</span><span class="sc">;</span>
132 119 <span class="s">}</span>
133 120 <span class="i">$DatabaseFingerprintsObject</span> = <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;GetFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
134 121 <span class="i">$DatabaseCmpdID</span> = <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;GetCompoundID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
135 122
136 123 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
137 124 <span class="i">@ComparisonValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
138 125 <span class="s">}</span>
139 126
140 127 <span class="j">REFERENCEFP:</span> <span class="k">for</span> <span class="i">$ReferenceIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceCmpdIDsRef</span>}}<span class="s">)</span> <span class="s">{</span>
141 128 <span class="i">$ReferenceCmpdID</span> = <span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceCmpdIDsRef</span>}-&gt;[<span class="i">$ReferenceIndex</span>]<span class="sc">;</span>
142 129 <span class="i">$ReferenceFingerprintsObject</span> = <span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceFingerprintsObjectsRef</span>}-&gt;[<span class="i">$ReferenceIndex</span>]<span class="sc">;</span>
143 130
144 131 <span class="i">$ComparisonValue</span> = <span class="i">CompareReferenceAndDatabaseFingerprintsPair</span><span class="s">(</span><span class="i">$ReferenceFingerprintsObject</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsObject</span><span class="s">)</span><span class="sc">;</span>
145 132 <span class="k">if</span> <span class="s">(</span>!<span class="k">defined</span> <span class="i">$ComparisonValue</span><span class="s">)</span> <span class="s">{</span>
146 133 <span class="k">next</span> <span class="j">REFERENCEFP</span><span class="sc">;</span>
147 134 <span class="s">}</span>
148 135
149 136 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>}<span class="s">)</span> <span class="s">{</span>
150 137 <span class="i">CollectSimilaritySearchResults</span><span class="s">(</span><span class="i">$DatabaseFingerprintsFileIO</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span><span class="sc">;</span>
151 138 <span class="s">}</span>
152 139 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
153 140 <span class="k">push</span> <span class="i">@ComparisonValues</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="sc">;</span>
154 141 <span class="s">}</span>
155 142 <span class="s">}</span>
156 143
157 144 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
158 145 <span class="i">$FusedComparisonValue</span> = <span class="i">CalculateGroupFusionComparisonValue</span><span class="s">(</span>\<span class="i">@ComparisonValues</span><span class="s">)</span><span class="sc">;</span>
159 146 <span class="k">if</span> <span class="s">(</span>!<span class="k">defined</span> <span class="i">$FusedComparisonValue</span><span class="s">)</span> <span class="s">{</span>
160 147 <span class="k">next</span> <span class="j">DATABASEFP</span><span class="sc">;</span>
161 148 <span class="s">}</span>
162 149 <span class="i">CollectSimilaritySearchResults</span><span class="s">(</span><span class="i">$DatabaseFingerprintsFileIO</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$FusedComparisonValue</span><span class="s">)</span><span class="sc">;</span>
163 150 <span class="s">}</span>
164 151 <span class="s">}</span>
165 152 <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
166 153
167 154 <span class="k">print</span> <span class="q">&quot;Number of fingerprints data entries in database fingerprints file: $FingerprintsCount\n&quot;</span><span class="sc">;</span>
168 155 <span class="k">print</span> <span class="q">&quot;Number of fingerprints date entries processed successfully: &quot;</span><span class="cm">,</span> <span class="s">(</span><span class="i">$FingerprintsCount</span> - <span class="i">$IgnoredFingerprintsCount</span><span class="s">)</span> <span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
169 156 <span class="k">print</span> <span class="q">&quot;Number of fingerprints data entries ignored due to missing/invalid data: $IgnoredFingerprintsCount\n\n&quot;</span><span class="sc">;</span>
170 157 <span class="s">}</span>
171 158
172 159 <span class="c"># Compare a pair of reference and database fingerprints objects corresponding to bit-vector or</span>
173 160 <span class="c"># vectors using specified comparison method and comparison cutoff...</span>
174 161 <span class="c">#</span>
175 <a name="CompareReferenceAndDatabaseFingerprintsPair-"></a> 162 <span class="k">sub </span><span class="m">CompareReferenceAndDatabaseFingerprintsPair</span> <span class="s">{</span>
176 163 <span class="k">my</span><span class="s">(</span><span class="i">$ReferenceFingerprintsObject</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsObject</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
177 164 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMethod</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="s">)</span><span class="sc">;</span>
178 165
179 166 <span class="i">$ComparisonMethod</span> = <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethod</span>}<span class="sc">;</span>
180 167 <span class="i">$ComparisonValue</span> = <span class="i">$ReferenceFingerprintsObject</span><span class="i">-&gt;$ComparisonMethod</span><span class="s">(</span><span class="i">$DatabaseFingerprintsObject</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethodParameters</span>}}<span class="s">)</span><span class="sc">;</span>
181 168
182 169 <span class="k">if</span> <span class="s">(</span>!<span class="k">defined</span> <span class="i">$ComparisonValue</span><span class="s">)</span> <span class="s">{</span>
183 170 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring fingerprints data for reference compound ID &quot;</span><span class="cm">,</span> <span class="i">$ReferenceFingerprintsObject</span><span class="i">-&gt;GetID</span><span class="s">(</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;: Its comparison with database compound ID, &quot;</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsObject</span><span class="i">-&gt;GetID</span><span class="s">(</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;, failed.\n&quot;</span><span class="sc">;</span>
184 171 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
185 172 <span class="s">}</span>
186 173
187 174 <span class="i">$ComparisonValue</span> = <span class="k">sprintf</span><span class="s">(</span><span class="q">&quot;%.$OptionsInfo{Precision}f&quot;</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="s">)</span><span class="sc">;</span>
188 175
189 176 <span class="c"># Apply any comparison cutoff...</span>
190 177 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyComparisonCutoff</span>}<span class="s">)</span> <span class="s">{</span>
191 178 <span class="k">return</span> <span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} ? <span class="s">(</span><span class="i">$ComparisonValue</span> &gt;= <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonCutoff</span>} ? <span class="i">$ComparisonValue</span> <span class="co">:</span> <span class="k">undef</span><span class="s">)</span> <span class="co">:</span> <span class="s">(</span><span class="i">$ComparisonValue</span> &lt;= <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonCutoff</span>} ? <span class="i">$ComparisonValue</span> <span class="co">:</span> <span class="k">undef</span><span class="s">)</span><span class="sc">;</span>
192 179 <span class="s">}</span>
193 180 <span class="k">else</span> <span class="s">{</span>
194 181 <span class="k">return</span> <span class="i">$ComparisonValue</span><span class="sc">;</span>
195 182 <span class="s">}</span>
196 183 <span class="s">}</span>
197 184
198 185 <span class="c"># Calculate group fusion comparison value...</span>
199 186 <span class="c">#</span>
200 <a name="CalculateGroupFusionComparisonValue-"></a> 187 <span class="k">sub </span><span class="m">CalculateGroupFusionComparisonValue</span> <span class="s">{</span>
201 188 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
202 189 <span class="k">my</span><span class="s">(</span><span class="i">$FusedComparisonValue</span><span class="cm">,</span> <span class="i">@ComparisonValues</span><span class="s">)</span><span class="sc">;</span>
203 190
204 191 <span class="k">if</span> <span class="s">(</span>!<span class="i">@</span>{<span class="i">$ComparisonValuesRef</span>}<span class="s">)</span> <span class="s">{</span>
205 192 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
206 193 <span class="s">}</span>
207 194
208 195 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">SortComparisonValues</span>}<span class="s">)</span> <span class="s">{</span>
209 196 <span class="i">@ComparisonValues</span> = <span class="k">sort</span> <span class="s">{</span> <span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} ? <span class="s">(</span><span class="i">$b</span> &lt;=&gt; <span class="i">$a</span><span class="s">)</span> <span class="co">:</span> <span class="s">(</span><span class="i">$a</span> &lt;=&gt; <span class="i">$b</span><span class="s">)</span> <span class="s">}</span> <span class="i">@</span>{<span class="i">$ComparisonValuesRef</span>}<span class="sc">;</span>
210 197 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">UsekNN</span>} &amp;&amp; <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">kNN</span>} &lt; <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$ComparisonValuesRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
211 198 <span class="c"># Keep only top kNN values for group fusion...</span>
212 199 <span class="k">splice</span> <span class="i">@ComparisonValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">kNN</span>}<span class="sc">;</span>
213 200 <span class="s">}</span>
214 201 <span class="i">$ComparisonValuesRef</span> = \<span class="i">@ComparisonValues</span><span class="sc">;</span>
215 202 <span class="s">}</span>
216 203
217 204 <span class="i">$FusedComparisonValue</span> = <span class="i">&amp;</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>}}<span class="s">(</span><span class="i">$ComparisonValuesRef</span><span class="s">)</span><span class="sc">;</span>
218 205 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>}<span class="s">)</span> <span class="s">{</span>
219 206 <span class="i">$FusedComparisonValue</span> = <span class="k">sprintf</span><span class="s">(</span><span class="q">&quot;%.$OptionsInfo{Precision}f&quot;</span><span class="cm">,</span> <span class="i">$FusedComparisonValue</span><span class="s">)</span><span class="sc">;</span>
220 207 <span class="s">}</span>
221 208
222 209 <span class="k">return</span> <span class="i">$FusedComparisonValue</span><span class="sc">;</span>
223 210 <span class="s">}</span>
224 211
225 212 <span class="c"># Collect similarity results for individual reference and multiple references search...</span>
226 213 <span class="c">#</span>
227 <a name="CollectSimilaritySearchResults-"></a> 214 <span class="k">sub </span><span class="m">CollectSimilaritySearchResults</span> <span class="s">{</span>
228 215 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseFingerprintsFileIO</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
229 216
230 217 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span> <span class="s">{</span>
231 218 <span class="i">$SimilaritySearchResults</span>{<span class="i">$ReferenceCmpdID</span>}<span class="i">-&gt;AddKeyValuePair</span><span class="s">(</span><span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="s">)</span><span class="sc">;</span>
232 219 <span class="s">}</span>
233 220 <span class="k">else</span> <span class="s">{</span>
234 221 <span class="i">$SimilaritySearchResults</span>{<span class="w">ResultsPseudoHeap</span>}<span class="i">-&gt;AddKeyValuePair</span><span class="s">(</span><span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="s">)</span><span class="sc">;</span>
235 222 <span class="s">}</span>
236 223
237 224 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectInputFileData</span>}<span class="s">)</span> <span class="s">{</span>
238 225 <span class="i">CollectDatabaseFileData</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
239 226 <span class="s">}</span>
240 227 <span class="s">}</span>
241 228
242 229 <span class="c"># Initialize similarity results for individual or multiple reference molecules...</span>
243 230 <span class="c">#</span>
244 <a name="InitializeSimilaritySearchResults-"></a> 231 <span class="k">sub </span><span class="m">InitializeSimilaritySearchResults</span> <span class="s">{</span>
245 232 <span class="k">my</span><span class="s">(</span><span class="i">$ReferenceCmpdID</span><span class="s">)</span><span class="sc">;</span>
246 233
247 234 <span class="i">%SimilaritySearchResults</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
248 235
249 236 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>}<span class="s">)</span> <span class="s">{</span>
250 237 <span class="k">for</span> <span class="i">$ReferenceCmpdID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceCmpdIDsRef</span>}}<span class="s">)</span> <span class="s">{</span>
251 238 <span class="i">$SimilaritySearchResults</span>{<span class="i">$ReferenceCmpdID</span>} = <span class="w">new</span> <span class="i">PseudoHeap</span><span class="s">(</span><span class="q">&#39;Type&#39;</span> <span class="cm">=&gt;</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} ? <span class="q">&#39;KeepTopN&#39;</span> <span class="co">:</span> <span class="q">&#39;KeepBottomN&#39;</span><span class="s">)</span><span class="cm">,</span> <span class="q">&#39;KeyType&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Numeric&#39;</span><span class="cm">,</span> <span class="q">&#39;MaxSize&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">MaxSimilarMolecules</span>}<span class="s">)</span><span class="sc">;</span>
252 239 <span class="s">}</span>
253 240 <span class="s">}</span>
254 241 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
255 242 <span class="i">$SimilaritySearchResults</span>{<span class="w">ResultsPseudoHeap</span>} = <span class="w">new</span> <span class="i">PseudoHeap</span><span class="s">(</span><span class="q">&#39;Type&#39;</span> <span class="cm">=&gt;</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} ? <span class="q">&#39;KeepTopN&#39;</span> <span class="co">:</span> <span class="q">&#39;KeepBottomN&#39;</span><span class="s">)</span><span class="cm">,</span> <span class="q">&#39;KeyType&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Numeric&#39;</span><span class="cm">,</span> <span class="q">&#39;MaxSize&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">MaxSimilarMolecules</span>}<span class="s">)</span><span class="sc">;</span>
256 243 <span class="s">}</span>
257 244
258 245 <span class="i">%DatabaseFingerprintsFileData</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
259 246 <span class="s">}</span>
260 247
261 248 <span class="c"># Write out results SD and/or CSV/TSV text files for individual or multiple reference molecules...</span>
262 249 <span class="c">#</span>
263 <a name="WriteSimilaritySearchResultFiles-"></a> 250 <span class="k">sub </span><span class="m">WriteSimilaritySearchResultFiles</span> <span class="s">{</span>
264 251 <span class="k">my</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="s">)</span><span class="sc">;</span>
265 252
266 253 <span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">SetupAndOpenOutputFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
267 254
268 255 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>}<span class="s">)</span> <span class="s">{</span>
269 256 <span class="k">for</span> <span class="i">$ReferenceCmpdID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceCmpdIDsRef</span>}}<span class="s">)</span> <span class="s">{</span>
270 257 <span class="k">for</span> <span class="i">$ComparisonValue</span> <span class="s">(</span><span class="i">$SimilaritySearchResults</span>{<span class="i">$ReferenceCmpdID</span>}<span class="i">-&gt;GetSortedKeys</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
271 258 <span class="k">for</span> <span class="i">$DatabaseCmpdID</span> <span class="s">(</span><span class="i">$SimilaritySearchResults</span>{<span class="i">$ReferenceCmpdID</span>}<span class="i">-&gt;GetKeyValues</span><span class="s">(</span><span class="i">$ComparisonValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
272 259 <span class="i">WriteDataToOutputFiles</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span><span class="sc">;</span>
273 260 <span class="s">}</span>
274 261 <span class="s">}</span>
275 262 <span class="s">}</span>
276 263 <span class="s">}</span>
277 264 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
278 265 <span class="k">for</span> <span class="i">$ComparisonValue</span> <span class="s">(</span><span class="i">$SimilaritySearchResults</span>{<span class="w">ResultsPseudoHeap</span>}<span class="i">-&gt;GetSortedKeys</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
279 266 <span class="k">for</span> <span class="i">$DatabaseCmpdID</span> <span class="s">(</span><span class="i">$SimilaritySearchResults</span>{<span class="w">ResultsPseudoHeap</span>}<span class="i">-&gt;GetKeyValues</span><span class="s">(</span><span class="i">$ComparisonValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
280 267 <span class="i">WriteDataToOutputFiles</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="s">)</span><span class="sc">;</span>
281 268 <span class="s">}</span>
282 269 <span class="s">}</span>
283 270 <span class="s">}</span>
284 271
285 272 <span class="k">if</span> <span class="s">(</span><span class="i">$NewSDFileRef</span><span class="s">)</span> <span class="s">{</span>
286 273 <span class="k">close</span> <span class="i">$NewSDFileRef</span><span class="sc">;</span>
287 274 <span class="s">}</span>
288 275 <span class="k">if</span> <span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> <span class="s">{</span>
289 276 <span class="k">close</span> <span class="i">$NewTextFileRef</span><span class="sc">;</span>
290 277 <span class="s">}</span>
291 278 <span class="s">}</span>
292 279
293 280 <span class="c"># Write individual reference or multiple references similarity results along with any other data to output files...</span>
294 281 <span class="c">#</span>
295 <a name="WriteDataToOutputFiles-"></a> 282 <span class="k">sub </span><span class="m">WriteDataToOutputFiles</span> <span class="s">{</span>
296 283 <span class="k">my</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="cm">,</span> <span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
297 284
298 285 <span class="k">if</span> <span class="s">(</span><span class="i">$NewSDFileRef</span><span class="s">)</span> <span class="s">{</span>
299 286 <span class="i">WriteMolStringDataToSDOutputFile</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$NewSDFileRef</span><span class="s">)</span><span class="sc">;</span>
300 287 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span> <span class="s">{</span>
301 288 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;&gt; &lt;ReferenceCmpdID&gt;\n$ReferenceCmpdID\n\n&quot;</span><span class="sc">;</span>
302 289 <span class="s">}</span>
303 290 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;&gt; &lt;DatabaseCmpdID&gt;\n$DatabaseCmpdID\n\n&gt; &lt;ComparisonValue&gt;\n$ComparisonValue\n\n&quot;</span><span class="sc">;</span>
304 291 <span class="i">WriteDatabaseDataToSDOutputFile</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$NewSDFileRef</span><span class="s">)</span><span class="sc">;</span>
305 292 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;\$\$\$\$\n&quot;</span><span class="sc">;</span>
306 293 <span class="s">}</span>
307 294
308 295 <span class="k">if</span> <span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> <span class="s">{</span>
309 296 <span class="k">my</span><span class="s">(</span><span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
310 297
311 298 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
312 299 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$ReferenceCmpdID</span><span class="s">)</span> <span class="s">{</span>
313 300 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$ReferenceCmpdID</span><span class="sc">;</span>
314 301 <span class="s">}</span>
315 302 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$ComparisonValue</span><span class="s">)</span><span class="sc">;</span>
316 303
317 304 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>} || <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>}<span class="s">)</span> <span class="s">{</span>
318 305 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">RetrieveDatabaseDataForTextOutputFile</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="s">)</span><span class="sc">;</span>
319 306 <span class="s">}</span>
320 307 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
321 308 <span class="s">}</span>
322 309 <span class="s">}</span>
323 310
324 311 <span class="c"># Open output files...</span>
325 312 <span class="c">#</span>
326 <a name="SetupAndOpenOutputFiles-"></a> 313 <span class="k">sub </span><span class="m">SetupAndOpenOutputFiles</span> <span class="s">{</span>
327 314 <span class="k">my</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$NewSDFile</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="s">)</span><span class="sc">;</span>
328 315
329 316 <span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="s">(</span><span class="k">undef</span><span class="s">)</span> x <span class="n">2</span><span class="sc">;</span>
330 317
331 318 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
332 319 <span class="i">$NewSDFile</span> = <span class="i">$OutputFilesInfo</span>{<span class="w">SDOutFileName</span>}<span class="sc">;</span>
333 320 <span class="k">print</span> <span class="q">&quot;Generating SD file $NewSDFile...\n&quot;</span><span class="sc">;</span>
334 321 <span class="k">open</span> <span class="w">NEWSDFILE</span><span class="cm">,</span> <span class="q">&quot;&gt;$NewSDFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $NewSDFile: $! \n&quot;</span><span class="sc">;</span>
335 322 <span class="i">$NewSDFileRef</span> = \<span class="i">*NEWSDFILE</span><span class="sc">;</span>
336 323 <span class="s">}</span>
337 324
338 325 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
339 326 <span class="i">$NewTextFile</span> = <span class="i">$OutputFilesInfo</span>{<span class="w">TextOutFileName</span>}<span class="sc">;</span>
340 327 <span class="k">print</span> <span class="q">&quot;Generating text file $NewTextFile...\n&quot;</span><span class="sc">;</span>
341 328 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">&quot;&gt;$NewTextFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $NewTextFile: $! \n&quot;</span><span class="sc">;</span>
342 329 <span class="i">$NewTextFileRef</span> = \<span class="i">*NEWTEXTFILE</span><span class="sc">;</span>
343 330
344 331 <span class="i">WriteTextFileCoulmnLabels</span><span class="s">(</span>\<span class="i">*NEWTEXTFILE</span><span class="s">)</span><span class="sc">;</span>
345 332 <span class="s">}</span>
346 333
347 334 <span class="k">return</span> <span class="s">(</span><span class="i">$NewSDFileRef</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span><span class="sc">;</span>
348 335 <span class="s">}</span>
349 336
350 337 <span class="c"># Write out approriate column labels to text file...</span>
351 338 <span class="c">#</span>
352 <a name="WriteTextFileCoulmnLabels-"></a> 339 <span class="k">sub </span><span class="m">WriteTextFileCoulmnLabels</span> <span class="s">{</span>
353 340 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
354 341 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
355 342
356 343 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
357 344
358 345 <span class="k">if</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>}<span class="s">)</span> <span class="s">{</span>
359 346 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="q">qw(ReferenceCompoundID DatabaseCompoundID ComparisonValue)</span><span class="sc">;</span>
360 347 <span class="s">}</span>
361 348 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>}<span class="s">)</span> <span class="s">{</span>
362 349 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="q">qw(DatabaseCompoundID ComparisonValue)</span><span class="sc">;</span>
363 350 <span class="s">}</span>
364 351
365 352 <span class="c"># Add columns for other database fingerprints file data to be written to output file...</span>
366 353 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>}<span class="s">)</span> <span class="s">{</span>
367 354 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataFieldsToOutput</span>}}<span class="sc">;</span>
368 355 <span class="s">}</span>
369 356 <span class="k">elsif</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>}<span class="s">)</span> <span class="s">{</span>
370 357 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabelsToOutput</span>}}<span class="sc">;</span>
371 358 <span class="s">}</span>
372 359
373 360 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
374 361 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">&quot;$Line\n&quot;</span><span class="sc">;</span>
375 362 <span class="s">}</span>
376 363
377 364 <span class="c"># Write molecule string data to SD output file...</span>
378 365 <span class="c">#</span>
379 <a name="WriteMolStringDataToSDOutputFile-"></a> 366 <span class="k">sub </span><span class="m">WriteMolStringDataToSDOutputFile</span> <span class="s">{</span>
380 367 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$NewSDFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
381 368
382 369 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectCmpdStringData</span>}<span class="s">)</span> <span class="s">{</span>
383 370 <span class="k">my</span><span class="s">(</span><span class="i">$MolString</span><span class="s">)</span><span class="sc">;</span>
384 371
385 372 <span class="s">(</span><span class="i">$MolString</span><span class="s">)</span> = <span class="k">split</span> <span class="q">/M END/</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}<span class="sc">;</span>
386 373 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;$MolString\nM END\n&quot;</span><span class="sc">;</span>
387 374 <span class="s">}</span>
388 375 <span class="k">else</span> <span class="s">{</span>
389 376 <span class="c"># Just write out an empty molecule data string...</span>
390 377 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="i">SDFileUtil::GenerateEmptyCtabBlockLines</span><span class="s">(</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
391 378 <span class="s">}</span>
392 379 <span class="s">}</span>
393 380
394 381 <span class="c"># Write database data from SD or Text database file to SD output file...</span>
395 382 <span class="c">#</span>
396 <a name="WriteDatabaseDataToSDOutputFile-"></a> 383 <span class="k">sub </span><span class="m">WriteDatabaseDataToSDOutputFile</span> <span class="s">{</span>
397 384 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$NewSDFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
398 385
399 386 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>}<span class="s">)</span> <span class="s">{</span>
400 387 <span class="k">my</span><span class="s">(</span><span class="i">$DataFieldLabel</span><span class="cm">,</span> <span class="i">$DataFieldValue</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">%DataFieldLabelAndValues</span><span class="s">)</span><span class="sc">;</span>
401 388
402 389 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">/\n/</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}<span class="sc">;</span>
403 390 <span class="i">%DataFieldLabelAndValues</span> = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
404 391
405 392 <span class="k">for</span> <span class="i">$DataFieldLabel</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputCurrentDataFields</span>} ? <span class="i">GetCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span> <span class="co">:</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataFieldsToOutput</span>}}<span class="s">)</span> <span class="s">{</span>
406 393 <span class="i">$DataFieldValue</span> = <span class="k">exists</span> <span class="i">$DataFieldLabelAndValues</span>{<span class="i">$DataFieldLabel</span>} ? <span class="i">$DataFieldLabelAndValues</span>{<span class="i">$DataFieldLabel</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="sc">;</span>
407 394 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;&gt; &lt;$DataFieldLabel&gt;\n$DataFieldValue\n\n&quot;</span><span class="sc">;</span>
408 395 <span class="s">}</span>
409 396 <span class="s">}</span>
410 397 <span class="k">elsif</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>}<span class="s">)</span> <span class="s">{</span>
411 398 <span class="k">my</span><span class="s">(</span><span class="i">$DataColNum</span><span class="cm">,</span> <span class="i">$DataFieldLabel</span><span class="cm">,</span> <span class="i">$DataFieldValue</span><span class="s">)</span><span class="sc">;</span>
412 399
413 400 <span class="k">for</span> <span class="i">$DataColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumsToOutput</span>}}<span class="s">)</span> <span class="s">{</span>
414 401 <span class="i">$DataFieldLabel</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumToLabelMap</span>}{<span class="i">$DataColNum</span>}<span class="sc">;</span>
415 402 <span class="i">$DataFieldValue</span> = <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}-&gt;[<span class="i">$DataColNum</span>]<span class="sc">;</span>
416 403 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">&quot;&gt; &lt;$DataFieldLabel&gt;\n$DataFieldValue\n\n&quot;</span><span class="sc">;</span>
417 404 <span class="s">}</span>
418 405 <span class="s">}</span>
419 406 <span class="s">}</span>
420 407
421 408 <span class="c"># Retriebe database data from SD or Text database file for text output file...</span>
422 409 <span class="c">#</span>
423 <a name="RetrieveDatabaseDataForTextOutputFile-"></a> 410 <span class="k">sub </span><span class="m">RetrieveDatabaseDataForTextOutputFile</span> <span class="s">{</span>
424 411 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
425 412
426 413 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>}<span class="s">)</span> <span class="s">{</span>
427 414 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">%DataFieldLabelAndValues</span><span class="s">)</span><span class="sc">;</span>
428 415
429 416 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">/\n/</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}<span class="sc">;</span>
430 417 <span class="i">%DataFieldLabelAndValues</span> = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
431 418
432 419 <span class="k">return</span> <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValues</span>{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValues</span>{<span class="i">$_</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataFieldsToOutput</span>}}<span class="sc">;</span>
433 420 <span class="s">}</span>
434 421 <span class="k">elsif</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>}<span class="s">)</span> <span class="s">{</span>
435 422 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}<span class="s">)</span> <span class="s">{</span>
436 423 <span class="k">return</span> <span class="k">map</span> <span class="s">{</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}-&gt;[<span class="i">$_</span>] <span class="s">}</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumsToOutput</span>}}<span class="s">)</span><span class="sc">;</span>
437 424 <span class="s">}</span>
438 425 <span class="k">else</span> <span class="s">{</span>
439 426 <span class="k">return</span> <span class="s">(</span><span class="q">&#39;&#39;</span><span class="s">)</span> x <span class="i">$#</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumsToOutput</span>}}<span class="sc">;</span>
440 427 <span class="s">}</span>
441 428 <span class="s">}</span>
442 429 <span class="s">}</span>
443 430
444 431 <span class="c"># Collect database file SD compound string or CSV/TSV data line for generating results</span>
445 432 <span class="c"># files..</span>
446 433 <span class="c">#</span>
447 <a name="CollectDatabaseFileData-"></a> 434 <span class="k">sub </span><span class="m">CollectDatabaseFileData</span> <span class="s">{</span>
448 435 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseCmpdID</span><span class="cm">,</span> <span class="i">$DatabaseFingerprintsFileIO</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
449 436
450 437 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>}<span class="s">)</span> <span class="s">{</span>
451 438 <span class="k">return</span><span class="sc">;</span>
452 439 <span class="s">}</span>
453 440
454 441 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectCmpdStringData</span>}<span class="s">)</span> <span class="s">{</span>
455 442 <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>} = <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;GetCompoundString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
456 443 <span class="s">}</span>
457 444
458 445 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectDataLine</span>}<span class="s">)</span> <span class="s">{</span>
459 446 <span class="k">my</span><span class="s">(</span><span class="i">@DataLineWords</span><span class="s">)</span><span class="sc">;</span>
460 447 <span class="i">@DataLineWords</span> = <span class="i">$DatabaseFingerprintsFileIO</span><span class="i">-&gt;GetDataLineWords</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
461 448 <span class="i">$DatabaseFingerprintsFileData</span>{<span class="i">$DatabaseCmpdID</span>} = \<span class="i">@DataLineWords</span><span class="sc">;</span>
462 449 <span class="s">}</span>
463 450
464 451 <span class="s">}</span>
465 452
466 453 <span class="c"># Read fingerprints data from reference fingerprints file...</span>
467 454 <span class="c">#</span>
468 <a name="ReadReferenceFingerprintsData-"></a> 455 <span class="k">sub </span><span class="m">ReadReferenceFingerprintsData</span> <span class="s">{</span>
469 456 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
470 457
471 458 <span class="i">$FingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsFileIOParameters</span>}}<span class="s">)</span><span class="sc">;</span>
472 459 <span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceCmpdIDsRef</span>}<span class="cm">,</span> <span class="i">$SimilaritySearchInfo</span>{<span class="w">ReferenceFingerprintsObjectsRef</span>}<span class="s">)</span> = <span class="i">Fingerprints::FingerprintsFileUtil::ReadAndProcessFingerpritsData</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
473 460
474 461 <span class="s">}</span>
475 462
476 463 <span class="c"># Retrieve information about fingerprints files...</span>
477 464 <span class="c">#</span>
478 <a name="RetrieveFingerprintsFilesInfo-"></a> 465 <span class="k">sub </span><span class="m">RetrieveFingerprintsFilesInfo</span> <span class="s">{</span>
479 466
480 467 <span class="i">%FingerprintsFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
481 468 <span class="i">%OutputFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
482 469 <span class="i">%SimilaritySearchInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
483 470
484 471 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
485 472 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
486 473
487 474 <span class="c"># Set up reference and database file names...</span>
488 475 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FileName</span>} = <span class="i">$FingerprintsFilesList</span>[<span class="n">0</span>]<span class="sc">;</span>
489 476 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FileName</span>} = <span class="i">$FingerprintsFilesList</span>[<span class="n">1</span>]<span class="sc">;</span>
490 477
491 478 <span class="c"># Retrieve information about reference and database fingerprints file...</span>
492 479 <span class="i">RetrieveReferenceFingerprintsFileInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
493 480 <span class="i">RetrieveDatabaseFingerprintsFileInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
494 481
495 482 <span class="c"># Setup fingerprints comparison method and associated method parameters...</span>
496 483 <span class="i">SetupReferenceAndDatabaseFingerprintsComparisonInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
497 484
498 485 <span class="c"># Retrieve information for output files...</span>
499 486 <span class="i">RetrieveOutputFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
500 487 <span class="s">}</span>
501 488
502 489 <span class="c"># Setup refrerence and database fingerprints comparison method and associated method parameters...</span>
503 490 <span class="c">#</span>
504 <a name="SetupReferenceAndDatabaseFingerprintsComparisonInfo-"></a> 491 <span class="k">sub </span><span class="m">SetupReferenceAndDatabaseFingerprintsComparisonInfo</span> <span class="s">{</span>
505 492
506 493 <span class="c"># Make sure reference and database fingerprints string match...</span>
507 494 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FirstFingerprintsStringType</span>} !~ <span class="q">/^$FingerprintsFilesInfo{Database}{FirstFingerprintsStringType}$/i</span><span class="s">)</span> ||
508 495 <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsBitVectorStringMode</span>} != <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsBitVectorStringMode</span>}<span class="s">)</span> ||
509 496 <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsVectorStringMode</span>} != <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsVectorStringMode</span>}<span class="s">)</span> <span class="s">)</span> <span class="s">{</span>
510 497 <span class="k">die</span> <span class="q">&quot;Error: First reference fingerprints string type, $FingerprintsFilesInfo{Reference}{FirstFingerprintsStringType}, must match first database fingerprints type, $FingerprintsFilesInfo{Database}{FirstFingerprintsStringType}.\n&quot;</span><span class="sc">;</span>
511 498 <span class="s">}</span>
512 499
513 500 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FirstFingerprintsStringDescription</span>} !~ <span class="q">/^$FingerprintsFilesInfo{Database}{FirstFingerprintsStringDescription}$/i</span><span class="s">)</span> <span class="s">{</span>
514 501 <span class="k">warn</span> <span class="q">&quot;Warning: First reference fingerprints string description, $FingerprintsFilesInfo{Reference}{FirstFingerprintsStringDescription}, doesn&#39;t match first database fingerprints string description, $FingerprintsFilesInfo{Database}{FirstFingerprintsStringDescription}.\n&quot;</span><span class="sc">;</span>
515 502 <span class="s">}</span>
516 503
517 504 <span class="c"># Setup individual reference and multiple references search mode...</span>
518 505 <span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>} = <span class="k">undef</span><span class="sc">;</span>
519 506 <span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>} = <span class="k">undef</span><span class="sc">;</span>
520 507
521 508 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^IndividualReference$/i</span><span class="s">)</span> <span class="s">{</span>
522 509 <span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>} = <span class="n">1</span><span class="sc">;</span>
523 510 <span class="s">}</span>
524 511 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^MultipleReferences$/i</span><span class="s">)</span> <span class="s">{</span>
525 512 <span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>} = <span class="n">1</span><span class="sc">;</span>
526 513 <span class="s">}</span>
527 514 <span class="k">else</span> <span class="s">{</span>
528 515 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{mode}, for option \&quot;-m, --mode\&quot; is not valid. Allowed values: IndividualReference, MultipleReferences\n&quot;</span><span class="sc">;</span>
529 516 <span class="s">}</span>
530 517
531 518 <span class="c"># Set up reference and database fingerprints similarity search method and paramaters...</span>
532 519 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$ComparisonMethod</span><span class="cm">,</span> <span class="i">$ApplyComparisonCutoff</span><span class="cm">,</span> <span class="i">$ComparisonCutoff</span><span class="cm">,</span> <span class="i">$KeepTop</span><span class="cm">,</span> <span class="i">@ComparisonMethodParameters</span><span class="s">)</span><span class="sc">;</span>
533 520
534 521 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethod</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
535 522 <span class="i">@</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethodParameters</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
536 523
537 524 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonCutoff</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
538 525 <span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
539 526
540 527 <span class="i">$ComparisonMeasure</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span> <span class="i">$ComparisonMethod</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
541 528 <span class="i">@ComparisonMethodParameters</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
542 529
543 530 <span class="j">FINGERPRINTSTYPE:</span> <span class="s">{</span>
544 531 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsBitVectorStringMode</span>}<span class="s">)</span> <span class="s">{</span>
545 532 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonMeasure</span>}<span class="sc">;</span>
546 533 <span class="i">$ComparisonMethod</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonMeasureMethod</span>}<span class="sc">;</span>
547 534
548 535 <span class="k">if</span> <span class="s">(</span><span class="i">$ComparisonMeasure</span> =~ <span class="q">/^TverskySimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
549 536 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>}<span class="sc">;</span>
550 537 <span class="s">}</span>
551 538 <span class="k">elsif</span> <span class="s">(</span><span class="i">$ComparisonMeasure</span> =~ <span class="q">/^WeightedTverskySimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
552 539 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>}<span class="sc">;</span>
553 540 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>}<span class="sc">;</span>
554 541 <span class="s">}</span>
555 542 <span class="k">elsif</span> <span class="s">(</span><span class="i">$ComparisonMeasure</span> =~ <span class="q">/^WeightedTanimotoSimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
556 543 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>}<span class="sc">;</span>
557 544 <span class="s">}</span>
558 545
559 546 <span class="k">last</span> <span class="j">FINGERPRINTSTYPE</span><span class="sc">;</span>
560 547 <span class="s">}</span>
561 548 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsVectorStringMode</span>}<span class="s">)</span> <span class="s">{</span>
562 549 <span class="k">my</span><span class="s">(</span><span class="i">$SkipValuesCheck</span><span class="s">)</span><span class="sc">;</span>
563 550
564 551 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMeasure</span>}<span class="sc">;</span>
565 552 <span class="i">$ComparisonMethod</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMeasuresMethod</span>}<span class="sc">;</span>
566 553
567 554 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMode</span>}<span class="sc">;</span>
568 555
569 556 <span class="i">$SkipValuesCheck</span> = <span class="i">$OptionsInfo</span>{<span class="w">Fast</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
570 557 <span class="k">push</span> <span class="i">@ComparisonMethodParameters</span><span class="cm">,</span> <span class="i">$SkipValuesCheck</span><span class="sc">;</span>
571 558
572 559 <span class="k">last</span> <span class="j">FINGERPRINTSTYPE</span><span class="sc">;</span>
573 560 <span class="s">}</span>
574 561 <span class="k">die</span> <span class="q">&quot;Error: Uknown fingerprints string type. Supported values: FingerprintsBitVectorString or FingerprintsVectorString.\n&quot;</span><span class="sc">;</span>
575 562 <span class="s">}</span>
576 563
577 564 <span class="i">$ApplyComparisonCutoff</span> = <span class="i">$SimilaritySearchInfo</span>{<span class="w">IndividualReferenceMode</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$SimilaritySearchInfo</span>{<span class="w">MultipleReferencesMode</span>} &amp;&amp; <span class="i">$OptionsInfo</span>{<span class="w">GroupFusionApplyCutoff</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="s">)</span><span class="sc">;</span>
578 565
579 566 <span class="i">$ComparisonCutoff</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span> <span class="i">$KeepTop</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
580 567 <span class="k">if</span> <span class="s">(</span><span class="i">$ComparisonMethod</span> =~ <span class="q">/Distance/i</span><span class="s">)</span> <span class="s">{</span>
581 568 <span class="i">$ComparisonCutoff</span> = <span class="i">$OptionsInfo</span>{<span class="w">DistanceCutoff</span>}<span class="sc">;</span>
582 569 <span class="i">$KeepTop</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SearchMode</span>} =~ <span class="q">/^SimilaritySearch$/i</span><span class="s">)</span> ? <span class="n">0</span> <span class="co">:</span> <span class="n">1</span><span class="sc">;</span>
583 570 <span class="s">}</span>
584 571 <span class="k">else</span> <span class="s">{</span>
585 572 <span class="i">$ComparisonCutoff</span> = <span class="i">$OptionsInfo</span>{<span class="w">SimilarityCutoff</span>}<span class="sc">;</span>
586 573 <span class="i">$KeepTop</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SearchMode</span>} =~ <span class="q">/^SimilaritySearch$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
587 574 <span class="s">}</span>
588 575
589 576 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethod</span>} = <span class="i">$ComparisonMethod</span><span class="sc">;</span>
590 577 <span class="i">@</span>{<span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonMethodParameters</span>}} = <span class="i">@ComparisonMethodParameters</span><span class="sc">;</span>
591 578
592 579 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ComparisonCutoff</span>} = <span class="i">$ComparisonCutoff</span><span class="sc">;</span>
593 580 <span class="i">$SimilaritySearchInfo</span>{<span class="w">KeepTop</span>} = <span class="i">$KeepTop</span><span class="sc">;</span>
594 581 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyComparisonCutoff</span>} = <span class="i">$ApplyComparisonCutoff</span><span class="sc">;</span>
595 582
596 583 <span class="c"># Setup references to group fusion methods...</span>
597 584 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = <span class="k">undef</span><span class="sc">;</span>
598 585 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>} = <span class="k">undef</span><span class="sc">;</span>
599 586
600 587 <span class="j">FUSIONRULE:</span> <span class="s">{</span>
601 588 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Max$/i</span><span class="s">)</span> <span class="s">{</span>
602 589 <span class="c"># It&#39;s always the first value in the appropriated sorted list using value of KeepTop...</span>
603 590 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = <span class="k">sub</span> <span class="s">{</span> <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> <span class="k">return</span> <span class="i">$ComparisonValuesRef</span>-&gt;[<span class="n">0</span>]<span class="sc">;</span> <span class="s">}</span><span class="sc">;</span>
604 591 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
605 592 <span class="s">}</span>
606 593 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Min$/i</span><span class="s">)</span> <span class="s">{</span>
607 594 <span class="c"># It&#39;s always the last value in the appropriated sorted list using value of KeepTop...</span>
608 595 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = <span class="k">sub</span> <span class="s">{</span> <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> <span class="k">return</span> <span class="i">$ComparisonValuesRef</span>-&gt;[<span class="i">$#</span>{<span class="i">$ComparisonValuesRef</span>}]<span class="sc">;</span> <span class="s">}</span><span class="sc">;</span>
609 596 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
610 597 <span class="s">}</span>
611 598 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Mean$/i</span><span class="s">)</span> <span class="s">{</span>
612 599 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = \<span class="i">&amp;StatisticsUtil::Mean</span><span class="sc">;</span>
613 600 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>} = <span class="n">1</span><span class="sc">;</span>
614 601 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
615 602 <span class="s">}</span>
616 603 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Median$/i</span><span class="s">)</span> <span class="s">{</span>
617 604 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = \<span class="i">&amp;StatisticsUtil::Median</span><span class="sc">;</span>
618 605 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>} = <span class="n">1</span><span class="sc">;</span>
619 606 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
620 607 <span class="s">}</span>
621 608 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Sum$/i</span><span class="s">)</span> <span class="s">{</span>
622 609 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = \<span class="i">&amp;StatisticsUtil::Sum</span><span class="sc">;</span>
623 610 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>} = <span class="n">1</span><span class="sc">;</span>
624 611 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
625 612 <span class="s">}</span>
626 613 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^Euclidean$/i</span><span class="s">)</span> <span class="s">{</span>
627 614 <span class="i">$SimilaritySearchInfo</span>{<span class="w">GroupFusionMethodRef</span>} = \<span class="i">&amp;StatisticsUtil::Euclidean</span><span class="sc">;</span>
628 615 <span class="i">$SimilaritySearchInfo</span>{<span class="w">ApplyPrecisionDuringFusion</span>} = <span class="n">1</span><span class="sc">;</span>
629 616 <span class="k">last</span> <span class="j">FUSIONRULE</span><span class="sc">;</span>
630 617 <span class="s">}</span>
631 618 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{groupfusionrule}, for option \&quot;-g, --GroupFusionRule\&quot; is not valid. Allowed values: Max, Min, Mean, Median, Sum, Euclidean\n&quot;</span><span class="sc">;</span>
632 619 <span class="s">}</span>
633 620
634 621 <span class="i">$SimilaritySearchInfo</span>{<span class="w">UsekNN</span>} = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">kNN</span>} !~ <span class="q">/^All$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
635 622 <span class="i">$SimilaritySearchInfo</span>{<span class="w">SortComparisonValues</span>} = <span class="s">(</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} =~ <span class="q">/^(Max|Min)$/i</span><span class="s">)</span> || <span class="i">$SimilaritySearchInfo</span>{<span class="w">UsekNN</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
636 623 <span class="s">}</span>
637 624
638 625 <span class="c"># Retrieve information about reference fingerprints file...</span>
639 626 <span class="c">#</span>
640 <a name="RetrieveReferenceFingerprintsFileInfo-"></a> 627 <span class="k">sub </span><span class="m">RetrieveReferenceFingerprintsFileInfo</span> <span class="s">{</span>
641 628 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$FingerprintsFileIO</span><span class="cm">,</span> <span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span><span class="sc">;</span>
642 629
643 630 <span class="i">$FingerprintsFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FileName</span>}<span class="sc">;</span>
644 631 <span class="s">(</span><span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span> = <span class="i">RetrieveFingerprintsFileInfo</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
645 632
646 633 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FileType</span>} = <span class="i">$FileType</span><span class="sc">;</span>
647 634 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">InDelim</span>} = <span class="i">$InDelim</span><span class="sc">;</span>
648 635
649 636 <span class="c"># Setup reference FingerprintsFileIO parameters...</span>
650 637 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsFileIOParameters</span>}} = <span class="i">RetrieveFingerprintsFileIOParameters</span><span class="s">(</span><span class="q">&#39;Reference&#39;</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
651 638
652 639 <span class="c"># Make sure reference fingerprints data file contains valid and retrieve fingerprints string mode information...</span>
653 640 <span class="s">(</span><span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span> = <span class="i">RetrieveFingerprintsFileFingerprintsStringInfo</span><span class="s">(</span><span class="q">&#39;Reference&#39;</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
654 641 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsStringMode</span>} = <span class="i">$FingerprintsStringMode</span><span class="sc">;</span>
655 642 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsBitVectorStringMode</span>} = <span class="i">$FingerprintsBitVectorStringMode</span><span class="sc">;</span>
656 643 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FingerprintsVectorStringMode</span>} = <span class="i">$FingerprintsVectorStringMode</span><span class="sc">;</span>
657 644 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FirstFingerprintsStringType</span>} = <span class="i">$FirstFingerprintsStringType</span><span class="sc">;</span>
658 645 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FirstFingerprintsStringDescription</span>} = <span class="i">$FirstFingerprintsStringDescription</span><span class="sc">;</span>
659 646
660 647 <span class="s">}</span>
661 648
662 649 <span class="c"># Retrieve information about database fingerprints file...</span>
663 650 <span class="c">#</span>
664 <a name="RetrieveDatabaseFingerprintsFileInfo-"></a> 651 <span class="k">sub </span><span class="m">RetrieveDatabaseFingerprintsFileInfo</span> <span class="s">{</span>
665 652 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$FingerprintsFileIO</span><span class="cm">,</span> <span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span><span class="sc">;</span>
666 653
667 654 <span class="i">$FingerprintsFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FileName</span>}<span class="sc">;</span>
668 655 <span class="s">(</span><span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span> = <span class="i">RetrieveFingerprintsFileInfo</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
669 656
670 657 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FileType</span>} = <span class="i">$FileType</span><span class="sc">;</span>
671 658 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">InDelim</span>} = <span class="i">$InDelim</span><span class="sc">;</span>
672 659
673 660 <span class="c"># Setup reference FingerprintsFileIO parameters...</span>
674 661 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsFileIOParameters</span>}} = <span class="i">RetrieveFingerprintsFileIOParameters</span><span class="s">(</span><span class="q">&#39;Database&#39;</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
675 662
676 663 <span class="c"># Make sure database fingerprints data file contains valid and retrieve fingerprints string mode information...</span>
677 664 <span class="s">(</span><span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span> = <span class="i">RetrieveFingerprintsFileFingerprintsStringInfo</span><span class="s">(</span><span class="q">&#39;Database&#39;</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
678 665 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsStringMode</span>} = <span class="i">$FingerprintsStringMode</span><span class="sc">;</span>
679 666 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsBitVectorStringMode</span>} = <span class="i">$FingerprintsBitVectorStringMode</span><span class="sc">;</span>
680 667 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FingerprintsVectorStringMode</span>} = <span class="i">$FingerprintsVectorStringMode</span><span class="sc">;</span>
681 668 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FirstFingerprintsStringType</span>} = <span class="i">$FirstFingerprintsStringType</span><span class="sc">;</span>
682 669 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FirstFingerprintsStringDescription</span>} = <span class="i">$FirstFingerprintsStringDescription</span><span class="sc">;</span>
683 670
684 671 <span class="c"># Retrieve database fingerprints data field information for output file...</span>
685 672 <span class="c">#</span>
686 673 <span class="i">RetrieveDatabaseFingerprintsDataFieldsInfo</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span><span class="sc">;</span>
687 674
688 675 <span class="c"># Retrieve database fingerprints text file data columns information for output file...</span>
689 676 <span class="c">#</span>
690 677 <span class="i">RetrieveDatabaseFingerprintsDataColsInfo</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span><span class="sc">;</span>
691 678
692 679 <span class="c"># Any need to collect database compound string or data line for generation of results files...</span>
693 680 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectCmpdStringData</span>} = <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^SD$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
694 681 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectDataLine</span>} = <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^Text$/i</span> &amp;&amp; <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataColsMode</span>} =~ <span class="q">/^(All|Specify)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
695 682 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectInputFileData</span>} = <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectCmpdStringData</span>} || <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CollectDataLine</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
696 683
697 684 <span class="c"># Set maximum number of similar compounds to find for individual reference of set of multiple</span>
698 685 <span class="c"># reference compounds...</span>
699 686 <span class="c">#</span>
700 687 <span class="i">SetMaximumSimilarMoleculesToRetrieve</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span><span class="sc">;</span>
701 688 <span class="s">}</span>
702 689
703 690 <span class="c"># Retrieve database fingerprints data field information...</span>
704 691 <span class="c">#</span>
705 <a name="RetrieveDatabaseFingerprintsDataFieldsInfo-"></a> 692 <span class="k">sub </span><span class="m">RetrieveDatabaseFingerprintsDataFieldsInfo</span> <span class="s">{</span>
706 693 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
707 694 <span class="k">my</span><span class="s">(</span><span class="i">$CollectDataFields</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$AllDataFieldsRef</span><span class="cm">,</span> <span class="i">$CommonDataFieldsRef</span><span class="cm">,</span> <span class="i">@DataFieldsToOutput</span><span class="s">)</span><span class="sc">;</span>
708 695
709 696 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>} = <span class="n">0</span><span class="sc">;</span>
710 697 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataFieldsToOutput</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
711 698
712 699 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputCurrentDataFields</span>} = <span class="n">0</span><span class="sc">;</span>
713 700
714 701 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">AllDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
715 702 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CommonDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
716 703 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">SpecifiedDatabaseDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
717 704
718 705 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> !~ <span class="q">/^SD$/i</span><span class="s">)</span> <span class="s">{</span>
719 706 <span class="k">return</span><span class="sc">;</span>
720 707 <span class="s">}</span>
721 708
722 709 <span class="c"># No need to go over SD file and collect data fields for SD file during All DatabaseDataFieldsMode as</span>
723 710 <span class="c"># they would be retrieved from database SD file compound string during generation of output files...</span>
724 711 <span class="c">#</span>
725 712 <span class="i">$CollectDataFields</span> = <span class="s">(</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} &amp;&amp; <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} =~ <span class="q">/^(All|Common)$/i</span><span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>} &amp;&amp; <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
726 713
727 714 <span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$AllDataFieldsRef</span><span class="cm">,</span> <span class="i">$CommonDataFieldsRef</span><span class="s">)</span> = <span class="s">(</span><span class="k">undef</span><span class="s">)</span> x <span class="n">2</span><span class="sc">;</span>
728 715
729 716 <span class="k">if</span> <span class="s">(</span><span class="i">$CollectDataFields</span><span class="s">)</span> <span class="s">{</span>
730 717 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">&quot;$FingerprintsFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $FingerprintsFile: $! \n&quot;</span><span class="sc">;</span>
731 718 <span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$AllDataFieldsRef</span><span class="cm">,</span> <span class="i">$CommonDataFieldsRef</span><span class="s">)</span> = <span class="i">GetAllAndCommonCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="sc">;</span>
732 719 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
733 720 <span class="s">}</span>
734 721
735 722 <span class="i">@DataFieldsToOutput</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
736 723 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
737 724 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$AllDataFieldsRef</span><span class="s">)</span> <span class="s">{</span>
738 725 <span class="k">push</span> <span class="i">@DataFieldsToOutput</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$AllDataFieldsRef</span>}<span class="sc">;</span>
739 726 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">AllDataFields</span>}}<span class="cm">,</span> <span class="i">@</span>{<span class="i">$AllDataFieldsRef</span>}<span class="sc">;</span>
740 727 <span class="s">}</span>
741 728 <span class="k">else</span> <span class="s">{</span>
742 729 <span class="c"># Retrieve and output data fields and values dynamically...</span>
743 730 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputCurrentDataFields</span>} = <span class="n">1</span><span class="sc">;</span>
744 731 <span class="s">}</span>
745 732 <span class="s">}</span>
746 733 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span> <span class="s">{</span>
747 734 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$CommonDataFieldsRef</span><span class="s">)</span> <span class="s">{</span>
748 735 <span class="k">push</span> <span class="i">@DataFieldsToOutput</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$CommonDataFieldsRef</span>}<span class="sc">;</span>
749 736 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">CommonDataFields</span>}}<span class="cm">,</span> <span class="i">@</span>{<span class="i">$CommonDataFieldsRef</span>}<span class="sc">;</span>
750 737 <span class="s">}</span>
751 738 <span class="s">}</span>
752 739 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
753 740 <span class="k">push</span> <span class="i">@DataFieldsToOutput</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataFields</span>}}<span class="sc">;</span>
754 741 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">SpecifiedDatabaseDataFields</span>}}<span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataFields</span>}}<span class="sc">;</span>
755 742 <span class="s">}</span>
756 743
757 744 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} !~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
758 745 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataFields</span>} = <span class="n">1</span><span class="sc">;</span>
759 746 <span class="s">}</span>
760 747
761 748 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataFieldsToOutput</span>}}<span class="cm">,</span> <span class="i">@DataFieldsToOutput</span><span class="sc">;</span>
762 749
763 750 <span class="s">}</span>
764 751
765 752 <span class="c"># Retrieve database fingerprints data columns information...</span>
766 753 <span class="c">#</span>
767 <a name="RetrieveDatabaseFingerprintsDataColsInfo-"></a> 754 <span class="k">sub </span><span class="m">RetrieveDatabaseFingerprintsDataColsInfo</span> <span class="s">{</span>
768 755 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
769 756 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">$NumOfCols</span><span class="cm">,</span> <span class="i">@DataColLabels</span><span class="cm">,</span> <span class="i">@DataColLabelsToOutput</span><span class="cm">,</span> <span class="i">@DataColNumsToOutput</span><span class="cm">,</span> <span class="i">%DataColLabelToNumMap</span><span class="cm">,</span> <span class="i">%DataColNumToLabelMap</span><span class="s">)</span><span class="sc">;</span>
770 757
771 758 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>} = <span class="n">0</span><span class="sc">;</span>
772 759
773 760 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
774 761 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabelToNumMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
775 762 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumToLabelMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
776 763
777 764 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumsToOutput</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
778 765 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabelsToOutput</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
779 766
780 767 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> !~ <span class="q">/^Text$/i</span><span class="s">)</span> <span class="s">{</span>
781 768 <span class="k">return</span><span class="sc">;</span>
782 769 <span class="s">}</span>
783 770
784 771 <span class="i">@DataColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
785 772 <span class="i">@DataColLabelsToOutput</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
786 773 <span class="i">@DataColNumsToOutput</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
787 774
788 775 <span class="i">%DataColLabelToNumMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
789 776 <span class="i">%DataColNumToLabelMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
790 777
791 778 <span class="c"># Get column label line...</span>
792 779 <span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">&quot;$FingerprintsFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $FingerprintsFile: $! \n&quot;</span><span class="sc">;</span>
793 780 <span class="i">$Line</span> = <span class="i">TextUtil::GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span>
794 781 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span>
795 782
796 783 <span class="i">$InDelim</span> = <span class="s">(</span><span class="i">$InDelim</span> =~ <span class="q">/^Tab$/i</span><span class="s">)</span> ? <span class="q">&quot;\t&quot;</span> <span class="co">:</span> <span class="s">(</span><span class="i">$InDelim</span> =~ <span class="q">/semicolon/i</span> ? <span class="q">&quot;\;&quot;</span> <span class="co">:</span> <span class="q">&quot;\,&quot;</span><span class="s">)</span><span class="sc">;</span>
797 784
798 785 <span class="i">@DataColLabels</span> = <span class="i">TextUtil::SplitWords</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span><span class="sc">;</span>
799 786 <span class="i">$NumOfCols</span> = <span class="k">scalar</span> <span class="i">@DataColLabels</span><span class="sc">;</span>
800 787
801 788 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#DataColLabels</span><span class="s">)</span> <span class="s">{</span>
802 789 <span class="i">$ColLabel</span> = <span class="i">$DataColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span>
803 790 <span class="i">$DataColLabelToNumMap</span>{<span class="i">$ColLabel</span>} = <span class="i">$ColNum</span><span class="sc">;</span>
804 791 <span class="i">$DataColNumToLabelMap</span>{<span class="i">$ColNum</span>} = <span class="i">$ColLabel</span><span class="sc">;</span>
805 792 <span class="s">}</span>
806 793
807 794 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataColsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
808 795 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseColMode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
809 796 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataCols</span>}}<span class="s">)</span> <span class="s">{</span>
810 797 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> &gt; <span class="i">$NumOfCols</span><span class="s">)</span> <span class="s">{</span>
811 798 <span class="k">die</span> <span class="q">&quot;Error: Column number, $ColNum, specified using \&quot;--DatabaseDataCols\&quot; is not valid: It must be &lt;= $NumOfCols\n&quot;</span><span class="sc">;</span>
812 799 <span class="s">}</span>
813 800 <span class="k">push</span> <span class="i">@DataColNumsToOutput</span><span class="cm">,</span> <span class="s">(</span><span class="i">$ColNum</span> - <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
814 801 <span class="s">}</span>
815 802 <span class="s">}</span>
816 803 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseColMode</span>} =~ <span class="q">/^ColLabel$/i</span><span class="s">)</span> <span class="s">{</span>
817 804 <span class="k">for</span> <span class="i">$ColLabel</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataCols</span>}}<span class="s">)</span> <span class="s">{</span>
818 805 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$DataColLabelToNumMap</span>{<span class="i">$ColLabel</span>}<span class="s">)</span> <span class="s">{</span>
819 806 <span class="k">die</span> <span class="q">&quot;Error: Column label, $ColLabel, specified using \&quot;--DatabaseDataCols\&quot; is not valid: It doesn&#39;t exist\n&quot;</span><span class="sc">;</span>
820 807 <span class="s">}</span>
821 808 <span class="k">push</span> <span class="i">@DataColNumsToOutput</span><span class="cm">,</span> <span class="i">$DataColLabelToNumMap</span>{<span class="i">$ColLabel</span>}<span class="sc">;</span>
822 809 <span class="s">}</span>
823 810 <span class="s">}</span>
824 811 <span class="s">}</span>
825 812 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataColsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
826 813 <span class="i">@DataColNumsToOutput</span> = <span class="k">map</span> <span class="s">{</span> <span class="i">$_</span> <span class="s">}</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#DataColLabels</span><span class="s">)</span><span class="sc">;</span>
827 814 <span class="s">}</span>
828 815
829 816 <span class="c"># Setup data column labels to output...</span>
830 817 <span class="k">if</span> <span class="s">(</span><span class="k">scalar</span> <span class="i">@DataColNumsToOutput</span><span class="s">)</span> <span class="s">{</span>
831 818 <span class="i">@DataColLabelsToOutput</span> = <span class="k">map</span> <span class="s">{</span> <span class="i">$DataColNumToLabelMap</span>{<span class="i">$_</span>} <span class="s">}</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#DataColNumsToOutput</span><span class="s">)</span><span class="sc">;</span>
832 819 <span class="s">}</span>
833 820
834 821 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">OutputDataCols</span>} = <span class="k">scalar</span> <span class="i">@DataColNumsToOutput</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
835 822
836 823 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabels</span>}} = <span class="i">@DataColLabels</span><span class="sc">;</span>
837 824 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabelToNumMap</span>}} = <span class="i">%DataColLabelToNumMap</span><span class="sc">;</span>
838 825 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumToLabelMap</span>}} = <span class="i">%DataColNumToLabelMap</span><span class="sc">;</span>
839 826
840 827 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColNumsToOutput</span>}} = <span class="i">@DataColNumsToOutput</span><span class="sc">;</span>
841 828 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">DataColLabelsToOutput</span>}} = <span class="i">@DataColLabelsToOutput</span><span class="sc">;</span>
842 829 <span class="s">}</span>
843 830
844 831 <span class="c"># Set maximum number of similar compounds to find for individual reference of set of multiple</span>
845 832 <span class="c"># reference compounds...</span>
846 833 <span class="c">#</span>
847 <a name="SetMaximumSimilarMoleculesToRetrieve-"></a> 834 <span class="k">sub </span><span class="m">SetMaximumSimilarMoleculesToRetrieve</span> <span class="s">{</span>
848 835 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
849 836 <span class="k">my</span><span class="s">(</span><span class="i">$MaxSimilarMolecules</span><span class="cm">,</span> <span class="i">$NumOfDatabaseMolecules</span><span class="cm">,</span> <span class="i">$PercentSimilarMolecules</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
850 837
851 838 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SimilarCountMode</span>} !~ <span class="q">/^PercentSimilar$/i</span><span class="s">)</span> <span class="s">{</span>
852 839 <span class="k">return</span><span class="sc">;</span>
853 840 <span class="s">}</span>
854 841
855 842 <span class="i">$PercentSimilarMolecules</span> = <span class="i">$OptionsInfo</span>{<span class="w">PercentSimilarMolecules</span>}<span class="sc">;</span>
856 843
857 844 <span class="c"># Count database entries to figure out MaxSimilarMolecules using PercentSimilarMolecules</span>
858 845 <span class="c"># value...</span>
859 846 <span class="i">$NumOfDatabaseMolecules</span> = <span class="n">0</span><span class="sc">;</span>
860 847 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^SD$/i</span> &amp;&amp; <span class="k">exists</span><span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">NumOfDatabaseMolecules</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
861 848 <span class="c"># It might already be counted for SD file...</span>
862 849 <span class="i">$NumOfDatabaseMolecules</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">NumOfDatabaseMolecules</span>}<span class="sc">;</span>
863 850 <span class="s">}</span>
864 851 <span class="k">else</span> <span class="s">{</span>
865 852 <span class="k">print</span> <span class="q">&quot;Calculating maximum number of similar molecules to retrieve for \&quot;PercentSimilar\&quot; value of \&quot;--SimilarCountMode\&quot; option by counting number of molecules in database fingerprints file...\n&quot;</span><span class="sc">;</span>
866 853 <span class="k">open</span> <span class="w">FINGERPRINTSFILE</span><span class="cm">,</span> <span class="q">&quot;$FingerprintsFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $FingerprintsFile: $! \n&quot;</span><span class="sc">;</span>
867 854 <span class="j">FILETYPE:</span> <span class="s">{</span>
868 855 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^SD$/i</span><span class="s">)</span> <span class="s">{</span>
869 856 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">TextUtil::GetTextLine</span><span class="s">(</span>\<span class="i">*FINGERPRINTSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
870 857 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^\$\$\$\$/</span><span class="s">)</span> <span class="s">{</span>
871 858 <span class="i">$NumOfDatabaseMolecules</span>++<span class="sc">;</span>
872 859 <span class="s">}</span>
873 860 <span class="s">}</span>
874 861 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
875 862 <span class="s">}</span>
876 863 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^Text$/i</span><span class="s">)</span> <span class="s">{</span>
877 864 <span class="c"># Ignore column label line...</span>
878 865 <span class="i">$Line</span> = <span class="i">TextUtil::GetTextLine</span><span class="s">(</span>\<span class="i">*FINGERPRINTSFILE</span><span class="s">)</span><span class="sc">;</span>
879 866 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">TextUtil::GetTextLine</span><span class="s">(</span>\<span class="i">*FINGERPRINTSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
880 867 <span class="i">$NumOfDatabaseMolecules</span>++<span class="sc">;</span>
881 868 <span class="s">}</span>
882 869 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
883 870 <span class="s">}</span>
884 871 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^FP$/i</span><span class="s">)</span> <span class="s">{</span>
885 872 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">TextUtil::GetTextLine</span><span class="s">(</span>\<span class="i">*FINGERPRINTSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
886 873 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> !~ <span class="q">/^#/</span><span class="s">)</span> <span class="s">{</span>
887 874 <span class="i">$NumOfDatabaseMolecules</span>++<span class="sc">;</span>
888 875 <span class="s">}</span>
889 876 <span class="s">}</span>
890 877 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
891 878 <span class="s">}</span>
892 879 <span class="i">$NumOfDatabaseMolecules</span> = <span class="n">0</span><span class="sc">;</span>
893 880 <span class="s">}</span>
894 881 <span class="k">close</span> <span class="w">FINGERPRINTSFILE</span><span class="sc">;</span>
895 882 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">NumOfDatabaseMolecules</span>} = <span class="i">$NumOfDatabaseMolecules</span><span class="sc">;</span>
896 883 <span class="s">}</span>
897 884
898 885 <span class="i">$MaxSimilarMolecules</span> = <span class="k">int</span> <span class="s">(</span><span class="s">(</span><span class="i">$NumOfDatabaseMolecules</span> * <span class="i">$PercentSimilarMolecules</span><span class="s">)</span>/<span class="n">100</span><span class="s">)</span><span class="sc">;</span>
899 886 <span class="k">if</span> <span class="s">(</span><span class="i">$MaxSimilarMolecules</span> &lt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
900 887 <span class="i">$MaxSimilarMolecules</span> = <span class="n">1</span><span class="sc">;</span>
901 888 <span class="s">}</span>
902 889
903 890 <span class="i">$OptionsInfo</span>{<span class="w">MaxSimilarMolecules</span>} = <span class="i">$MaxSimilarMolecules</span><span class="sc">;</span>
904 891 <span class="s">}</span>
905 892
906 893 <span class="c"># Retrieve information about fingerprints file...</span>
907 894 <span class="c">#</span>
908 <a name="RetrieveFingerprintsFileInfo-"></a> 895 <span class="k">sub </span><span class="m">RetrieveFingerprintsFileInfo</span> <span class="s">{</span>
909 896 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
910 897 <span class="k">my</span><span class="s">(</span><span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$FileName</span><span class="s">)</span><span class="sc">;</span>
911 898
912 899 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
913 900 <span class="k">die</span> <span class="q">&quot;Error: Input fingerprints file, $FingerprintsFile, doesn&#39;t exist.\n&quot;</span><span class="sc">;</span>
914 901 <span class="s">}</span>
915 902
916 903 <span class="i">$FileType</span> = <span class="i">Fingerprints::FingerprintsFileUtil::GetFingerprintsFileType</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
917 904 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$FileType</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
918 905 <span class="k">die</span> <span class="q">&quot;Error: Input file, $FingerprintsFile, is not a fingerprints file.\n&quot;</span><span class="sc">;</span>
919 906 <span class="s">}</span>
920 907
921 908 <span class="i">$InDelim</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
922 909 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^Text$/i</span><span class="s">)</span> <span class="s">{</span>
923 910 <span class="i">$FileDir</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span>
924 911 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
925 912 <span class="i">$InDelim</span> = <span class="s">(</span><span class="i">$FileExt</span> =~ <span class="q">/^tsv$/i</span><span class="s">)</span> ? <span class="q">&#39;Tab&#39;</span> <span class="co">:</span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="sc">;</span>
926 913 <span class="s">}</span>
927 914
928 915 <span class="k">return</span> <span class="s">(</span><span class="i">$FileType</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="s">)</span><span class="sc">;</span>
929 916 <span class="s">}</span>
930 917
931 918 <span class="c"># Retrieve fingerprints file IO parameters...</span>
932 919 <span class="c">#</span>
933 <a name="RetrieveFingerprintsFileIOParameters-"></a> 920 <span class="k">sub </span><span class="m">RetrieveFingerprintsFileIOParameters</span> <span class="s">{</span>
934 921 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileMode</span><span class="cm">,</span> <span class="i">$FileType</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
935 922 <span class="k">my</span><span class="s">(</span><span class="i">%FingerprintsFileIOParams</span><span class="s">)</span><span class="sc">;</span>
936 923
937 924 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFileMode</span> !~ <span class="q">/^(Reference|Database)$/</span><span class="s">)</span> <span class="s">{</span>
938 925 <span class="k">die</span> <span class="q">&quot;Error: Unknown fingerprints file mode: $FingerprintsFileMode. Supported values: Reference or Database\n&quot;</span><span class="sc">;</span>
939 926 <span class="s">}</span>
940 927
941 928 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
942 929
943 930 <span class="j">FILETYPE:</span> <span class="s">{</span>
944 931 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^SD$/i</span><span class="s">)</span> <span class="s">{</span>
945 932 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">&#39;Mode&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Read&#39;</span><span class="cm">,</span> <span class="q">&#39;FingerprintsStringMode&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsMode</span>}<span class="cm">,</span> <span class="q">&#39;ValidateData&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">&#39;DetailLevel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="cm">,</span> <span class="q">&#39;FingerprintsFieldLabel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}FingerprintsField&quot;</span>}<span class="cm">,</span> <span class="q">&#39;CompoundIDMode&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}CompoundIDMode&quot;</span>}<span class="cm">,</span> <span class="q">&#39;CompoundIDFieldLabel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}CompoundIDField&quot;</span>}<span class="cm">,</span> <span class="q">&#39;CompoundIDPrefix&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}CompoundIDPrefix&quot;</span>}<span class="s">)</span><span class="sc">;</span>
946 933 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
947 934 <span class="s">}</span>
948 935 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^FP$/i</span><span class="s">)</span> <span class="s">{</span>
949 936 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">&#39;Mode&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Read&#39;</span><span class="cm">,</span> <span class="q">&#39;FingerprintsStringMode&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsMode</span>}<span class="cm">,</span> <span class="q">&#39;ValidateData&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">&#39;DetailLevel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="s">)</span><span class="sc">;</span>
950 937 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
951 938 <span class="s">}</span>
952 939 <span class="k">if</span> <span class="s">(</span><span class="i">$FileType</span> =~ <span class="q">/^Text$/i</span><span class="s">)</span> <span class="s">{</span>
953 940 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">&#39;Mode&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Read&#39;</span><span class="cm">,</span> <span class="q">&#39;FingerprintsStringMode&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsMode</span>}<span class="cm">,</span> <span class="q">&#39;ValidateData&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">&#39;DetailLevel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="cm">,</span> <span class="q">&#39;FingerprintsCol&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}FingerprintsCol&quot;</span>}<span class="cm">,</span> <span class="q">&#39;ColMode&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}ColMode&quot;</span>}<span class="cm">,</span> <span class="q">&#39;CompoundIDCol&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}CompoundIDCol&quot;</span>}<span class="cm">,</span> <span class="q">&#39;CompoundIDPrefix&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="q">&quot;${FingerprintsFileMode}CompoundIDPrefix&quot;</span>}<span class="cm">,</span> <span class="q">&#39;InDelim&#39;</span> <span class="cm">=&gt;</span> <span class="i">$FingerprintsFilesInfo</span>{<span class="i">$FingerprintsFileMode</span>}{<span class="w">InDelim</span>}<span class="s">)</span><span class="sc">;</span>
954 941 <span class="k">last</span> <span class="j">FILETYPE</span><span class="sc">;</span>
955 942 <span class="s">}</span>
956 943 <span class="k">die</span> <span class="q">&quot;Error: Fingerprints file type, $FileType, is not valid. Supported file types: SD, FP or Text\n&quot;</span><span class="sc">;</span>
957 944 <span class="s">}</span>
958 945
959 946 <span class="k">return</span> <span class="i">%FingerprintsFileIOParams</span><span class="sc">;</span>
960 947 <span class="s">}</span>
961 948
962 949 <span class="c"># Make sure fingerprints data file contains valid dta and retrieve fingerprints string mode information...</span>
963 950 <span class="c">#</span>
964 <a name="RetrieveFingerprintsFileFingerprintsStringInfo-"></a> 951 <span class="k">sub </span><span class="m">RetrieveFingerprintsFileFingerprintsStringInfo</span> <span class="s">{</span>
965 952 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileMode</span><span class="cm">,</span> <span class="i">$FingerprintsFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
966 953 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="cm">,</span> <span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span><span class="sc">;</span>
967 954
968 955 <span class="i">$FingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="i">$FingerprintsFileMode</span>}{<span class="w">FingerprintsFileIOParameters</span>}}<span class="s">)</span><span class="sc">;</span>
969 956 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="s">)</span> <span class="s">{</span>
970 957 <span class="k">die</span> <span class="q">&quot;Error: Reference fingerprints file, $FingerprintsFile, contains invalid fingerprints data.\n&quot;</span><span class="sc">;</span>
971 958 <span class="s">}</span>
972 959 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="i">-&gt;IsFingerprintsFileDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
973 960 <span class="k">die</span> <span class="q">&quot;Error: Reference fingerprints file, $FingerprintsFile, contains invalid fingerprints data.\n&quot;</span><span class="sc">;</span>
974 961 <span class="s">}</span>
975 962
976 963 <span class="i">$FingerprintsStringMode</span> = <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;GetFingerprintsStringMode</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
977 964 <span class="i">$FingerprintsBitVectorStringMode</span> = <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;GetFingerprintsBitVectorStringMode</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
978 965 <span class="i">$FingerprintsVectorStringMode</span> = <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;GetFingerprintsVectorStringMode</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
979 966
980 967 <span class="i">$FirstFingerprintsStringType</span> = <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;GetFirstFingerprintsStringType</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
981 968 <span class="i">$FirstFingerprintsStringDescription</span> = <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;GetFirstFingerprintsStringDescription</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
982 969
983 970 <span class="i">$FingerprintsFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
984 971
985 972 <span class="k">return</span> <span class="s">(</span><span class="i">$FingerprintsStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsBitVectorStringMode</span><span class="cm">,</span> <span class="i">$FingerprintsVectorStringMode</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringType</span><span class="cm">,</span> <span class="i">$FirstFingerprintsStringDescription</span><span class="s">)</span><span class="sc">;</span>
986 973 <span class="s">}</span>
987 974
988 975 <span class="c"># Retrieve output files names using reference fingerprints file name...</span>
989 976 <span class="c">#</span>
990 <a name="RetrieveOutputFilesInfo-"></a> 977 <span class="k">sub </span><span class="m">RetrieveOutputFilesInfo</span> <span class="s">{</span>
991 978 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$SDOutFileName</span><span class="cm">,</span> <span class="i">$TextOutFileName</span><span class="cm">,</span> <span class="i">$SDOutFileExt</span><span class="cm">,</span> <span class="i">$TextOutFileExt</span><span class="cm">,</span> <span class="i">$ReferenceFileName</span><span class="cm">,</span> <span class="i">$DatabaseFileName</span><span class="s">)</span><span class="sc">;</span>
992 979
993 980 <span class="i">$OutputFilesInfo</span>{<span class="w">OutFileRoot</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
994 981 <span class="i">$OutputFilesInfo</span>{<span class="w">SDOutFileName</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
995 982 <span class="i">$OutputFilesInfo</span>{<span class="w">TextOutFileName</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
996 983
997 984 <span class="i">$FingerprintsFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FileName</span>}<span class="sc">;</span>
998 985
999 986 <span class="i">$FileDir</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span>
1000 987 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
1001 988
1002 989 <span class="i">$SDOutFileExt</span> = <span class="q">&quot;sdf&quot;</span><span class="sc">;</span>
1003 990 <span class="i">$TextOutFileExt</span> = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> ? <span class="q">&quot;tsv&quot;</span> <span class="co">:</span> <span class="q">&quot;csv&quot;</span><span class="sc">;</span>
1004 991
1005 992 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span> <span class="s">{</span>
1006 993 <span class="k">my</span> <span class="s">(</span><span class="i">$RootFileDir</span><span class="cm">,</span> <span class="i">$RootFileName</span><span class="cm">,</span> <span class="i">$RootFileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span>
1007 994 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> &amp;&amp; <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
1008 995 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
1009 996 <span class="s">}</span>
1010 997 <span class="k">else</span> <span class="s">{</span>
1011 998 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
1012 999 <span class="s">}</span>
1013 1000 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
1014 1001 <span class="s">}</span>
1015 1002 <span class="k">else</span> <span class="s">{</span>
1016 1003 <span class="i">$OutFileRoot</span> = <span class="q">&quot;${FileName}SimilaritySearching&quot;</span><span class="sc">;</span>
1017 1004 <span class="s">}</span>
1018 1005
1019 1006 <span class="i">$SDOutFileName</span> = <span class="q">&quot;${OutFileRoot}.${SDOutFileExt}&quot;</span><span class="sc">;</span>
1020 1007 <span class="i">$TextOutFileName</span> = <span class="q">&quot;${OutFileRoot}.${TextOutFileExt}&quot;</span><span class="sc">;</span>
1021 1008
1022 1009 <span class="i">$ReferenceFileName</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Reference</span>}{<span class="w">FileName</span>}<span class="sc">;</span>
1023 1010 <span class="i">$DatabaseFileName</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">Database</span>}{<span class="w">FileName</span>}<span class="sc">;</span>
1024 1011
1025 1012 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
1026 1013 <span class="k">if</span> <span class="s">(</span><span class="i">$SDOutFileName</span> =~ <span class="q">/^$ReferenceFileName$/i</span><span class="s">)</span> <span class="s">{</span>
1027 1014 <span class="k">die</span> <span class="q">&quot;Error: Same output, $SDOutFileName, and reference input file names.\nSpecify a different name using \&quot;-r --root\&quot; option or use default name.\n&quot;</span><span class="sc">;</span>
1028 1015 <span class="s">}</span>
1029 1016 <span class="k">if</span> <span class="s">(</span><span class="i">$SDOutFileName</span> =~ <span class="q">/^$DatabaseFileName$/i</span><span class="s">)</span> <span class="s">{</span>
1030 1017 <span class="k">die</span> <span class="q">&quot;Error: Same output, $SDOutFileName, and database input file names.\nSpecify a different name using \&quot;-r --root\&quot; option or use default name.\n&quot;</span><span class="sc">;</span>
1031 1018 <span class="s">}</span>
1032 1019 <span class="s">}</span>
1033 1020
1034 1021 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
1035 1022 <span class="k">if</span> <span class="s">(</span><span class="i">$TextOutFileName</span> =~ <span class="q">/^$ReferenceFileName$/i</span><span class="s">)</span> <span class="s">{</span>
1036 1023 <span class="k">die</span> <span class="q">&quot;Error: Same output, $TextOutFileName, and reference input file names.\nSpecify a different name using \&quot;-r --root\&quot; option or use default name.\n&quot;</span><span class="sc">;</span>
1037 1024 <span class="s">}</span>
1038 1025 <span class="k">if</span> <span class="s">(</span><span class="i">$TextOutFileName</span> =~ <span class="q">/^$DatabaseFileName$/i</span><span class="s">)</span> <span class="s">{</span>
1039 1026 <span class="k">die</span> <span class="q">&quot;Error: Same output, $TextOutFileName, and database input file names.\nSpecify a different name using \&quot;-r --root\&quot; option or use default name.\n&quot;</span><span class="sc">;</span>
1040 1027 <span class="s">}</span>
1041 1028 <span class="s">}</span>
1042 1029
1043 1030 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span>
1044 1031 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
1045 1032 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$SDOutFileName</span><span class="s">)</span> <span class="s">{</span>
1046 1033 <span class="k">die</span> <span class="q">&quot;Error: The output file $SDOutFileName already exists.\n&quot;</span><span class="sc">;</span>
1047 1034 <span class="s">}</span>
1048 1035 <span class="s">}</span>
1049 1036 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
1050 1037 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$TextOutFileName</span><span class="s">)</span> <span class="s">{</span>
1051 1038 <span class="k">die</span> <span class="q">&quot;Error: The output file $TextOutFileName already exists.\n&quot;</span><span class="sc">;</span>
1052 1039 <span class="s">}</span>
1053 1040 <span class="s">}</span>
1054 1041 <span class="s">}</span>
1055 1042
1056 1043 <span class="i">$OutputFilesInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$OutFileRoot</span><span class="sc">;</span>
1057 1044 <span class="i">$OutputFilesInfo</span>{<span class="w">SDOutFileName</span>} = <span class="i">$SDOutFileName</span><span class="sc">;</span>
1058 1045 <span class="i">$OutputFilesInfo</span>{<span class="w">TextOutFileName</span>} = <span class="i">$TextOutFileName</span><span class="sc">;</span>
1059 1046
1060 1047 <span class="s">}</span>
1061 1048
1062 1049 <span class="c"># Process input fingerprints file names...</span>
1063 1050 <span class="c">#</span>
1064 <a name="ProcessFingerprintsFileNames-"></a>1051 <span class="k">sub </span><span class="m">ProcessFingerprintsFileNames</span> <span class="s">{</span>
1065 1052 <span class="i">@FingerprintsFilesList</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1066 1053
1067 1054 <span class="k">if</span> <span class="s">(</span><span class="i">@ARGV</span> != <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
1068 1055 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">&quot;$FindBin::Bin/$ScriptName&quot;</span><span class="s">)</span><span class="sc">;</span>
1069 1056 <span class="s">}</span>
1070 1057
1071 1058 <span class="c"># Reference fingerprints file name...</span>
1072 1059 <span class="k">push</span> <span class="i">@FingerprintsFilesList</span><span class="cm">,</span> <span class="i">$ARGV</span>[<span class="n">0</span>]<span class="sc">;</span>
1073 1060
1074 1061 <span class="c"># Database fingerprints file name...</span>
1075 1062 <span class="k">push</span> <span class="i">@FingerprintsFilesList</span><span class="cm">,</span> <span class="i">$ARGV</span>[<span class="n">1</span>]<span class="sc">;</span>
1076 1063
1077 1064 <span class="s">}</span>
1078 1065
1079 1066 <span class="c"># Process option values...</span>
1080 <a name="ProcessOptions-"></a>1067 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
1081 1068 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1082 1069
1083 1070 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
1084 1071 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsMode</span>} = <span class="i">$Options</span>{<span class="w">fingerprintsmode</span>}<span class="sc">;</span>
1085 1072
1086 1073 <span class="i">$OptionsInfo</span>{<span class="w">SearchMode</span>} = <span class="i">$Options</span>{<span class="w">searchmode</span>}<span class="sc">;</span>
1087 1074
1088 1075 <span class="i">ProcessBitVectorComparisonOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1089 1076 <span class="i">ProcessVectorComparisonOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1090 1077
1091 1078 <span class="i">$OptionsInfo</span>{<span class="w">GroupFusionRule</span>} = <span class="i">$Options</span>{<span class="w">groupfusionrule</span>}<span class="sc">;</span>
1092 1079 <span class="i">$OptionsInfo</span>{<span class="w">GroupFusionApplyCutoff</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">groupfusionapplycutoff</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span><span class="sc">;</span>
1093 1080
1094 1081 <span class="i">$OptionsInfo</span>{<span class="w">SimilarCountMode</span>} = <span class="i">$Options</span>{<span class="w">similarcountmode</span>}<span class="sc">;</span>
1095 1082 <span class="i">$OptionsInfo</span>{<span class="w">NumOfSimilarMolecules</span>} = <span class="i">$Options</span>{<span class="w">numofsimilarmolecules</span>}<span class="sc">;</span>
1096 1083 <span class="i">$OptionsInfo</span>{<span class="w">PercentSimilarMolecules</span>} = <span class="i">$Options</span>{<span class="w">percentsimilarmolecules</span>}<span class="sc">;</span>
1097 1084
1098 1085 <span class="c"># Set MaxSimilarMolecules to NumOfSimilarMolecules. For PercentSimilar value of SimilarCountMode,</span>
1099 1086 <span class="c"># it&#39;ll be overwritten using number of entries in database fingerprints file and value of PercentSimilarMolecules...</span>
1100 1087 <span class="c">#</span>
1101 1088 <span class="i">$OptionsInfo</span>{<span class="w">MaxSimilarMolecules</span>} = <span class="i">$OptionsInfo</span>{<span class="w">NumOfSimilarMolecules</span>}<span class="sc">;</span>
1102 1089
1103 1090 <span class="i">$OptionsInfo</span>{<span class="w">SimilarityCutoff</span>} = <span class="i">$Options</span>{<span class="w">similaritycutoff</span>}<span class="sc">;</span>
1104 1091 <span class="i">$OptionsInfo</span>{<span class="w">DistanceCutoff</span>} = <span class="i">$Options</span>{<span class="w">distancecutoff</span>}<span class="sc">;</span>
1105 1092
1106 1093 <span class="i">$OptionsInfo</span>{<span class="w">kNN</span>} = <span class="i">$Options</span>{<span class="w">knn</span>}<span class="sc">;</span>
1107 1094 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">knn</span>} !~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
1108 1095 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">knn</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1109 1096 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{knn}, for option \&quot;-k, --KNN\&quot; is not valid. Allowed values: &gt; 0 \n&quot;</span><span class="sc">;</span>
1110 1097 <span class="s">}</span>
1111 1098 <span class="s">}</span>
1112 1099
1113 1100 <span class="i">ProcessReferenceFingerprintsDataOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1114 1101 <span class="i">ProcessDatabaseFingerprintsDataOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1115 1102
1116 1103 <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>} = <span class="i">$Options</span>{<span class="w">detail</span>}<span class="sc">;</span>
1117 1104
1118 1105 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="i">$Options</span>{<span class="w">indelim</span>}<span class="sc">;</span>
1119 1106 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/tab/i</span> <span class="s">)</span> ? <span class="q">&quot;\t&quot;</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span><span class="s">)</span> ? <span class="q">&quot;\;&quot;</span> <span class="co">:</span> <span class="q">&quot;\,&quot;</span><span class="s">)</span><span class="sc">;</span>
1120 1107 <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1121 1108
1122 1109 <span class="i">$OptionsInfo</span>{<span class="w">Output</span>} = <span class="i">$Options</span>{<span class="w">output</span>}<span class="sc">;</span>
1123 1110 <span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(SD|Both)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1124 1111 <span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(Text|Both)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1125 1112
1126 1113 <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1127 1114 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1128 1115
1129 1116 <span class="i">$OptionsInfo</span>{<span class="w">Fast</span>} = <span class="i">$Options</span>{<span class="w">fast</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
1130 1117 <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>} = <span class="i">$Options</span>{<span class="w">fast</span>} ? <span class="n">0</span> <span class="co">:</span> <span class="n">1</span><span class="sc">;</span>
1131 1118
1132 1119 <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>} = <span class="i">$Options</span>{<span class="w">precision</span>}<span class="sc">;</span>
1133 1120 <span class="s">}</span>
1134 1121
1135 1122 <span class="c"># Process options related to comparion of bit vector strings...</span>
1136 1123 <span class="c">#</span>
1137 <a name="ProcessBitVectorComparisonOptions-"></a>1124 <span class="k">sub </span><span class="m">ProcessBitVectorComparisonOptions</span> <span class="s">{</span>
1138 1125 <span class="c"># Setup supported bit vector similarity coefficients for bit vector strings...</span>
1139 1126 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$SupportedComparisonMeasure</span><span class="cm">,</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresNameMap</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresMethodMap</span><span class="s">)</span><span class="sc">;</span>
1140 1127
1141 1128 <span class="i">@SupportedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1142 1129 <span class="i">%SupportedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1143 1130 <span class="i">%SupportedComparisonMeasuresMethodMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1144 1131
1145 1132 <span class="k">for</span> <span class="i">$SupportedComparisonMeasure</span> <span class="s">(</span><span class="i">Fingerprints::FingerprintsBitVector::GetSupportedSimilarityCoefficients</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1146 1133 <span class="c"># Similarity coefficient function/method names contain &quot;Coefficient&quot; in their names.</span>
1147 1134 <span class="c"># So take &#39;em out and setup a map to original function/method name...</span>
1148 1135 <span class="i">$ComparisonMeasure</span> = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
1149 1136 <span class="i">$ComparisonMeasure</span> =~ <span class="q">s/Coefficient$//</span><span class="sc">;</span>
1150 1137
1151 1138 <span class="k">push</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
1152 1139 <span class="i">$SupportedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>} = <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
1153 1140 <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>} = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
1154 1141 <span class="s">}</span>
1155 1142
1156 1143 <span class="c"># Setup similarity coefficient to use for calculating similarity matrices for bit vector strings...</span>
1157 1144 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="cm">,</span> <span class="i">$SpecifiedComparisonMeasureName</span><span class="cm">,</span> <span class="i">$SpecifiedComparisonMeasureMethod</span><span class="s">)</span><span class="sc">;</span>
1158 1145
1159 1146 <span class="i">$SpecifiedComparisonMeasureName</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1160 1147 <span class="i">$SpecifiedComparisonMeasureMethod</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1161 1148
1162 1149 <span class="i">$SpecifiedMeasure</span> = <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>}<span class="sc">;</span>
1163 1150
1164 1151 <span class="k">if</span> <span class="s">(</span>! <span class="k">exists</span> <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>} <span class="s">)</span> <span class="s">{</span>
1165 1152 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $SpecifiedMeasure, for option \&quot;-b --BitVectorComparisonMode\&quot; is not valid.\nAllowed values:&quot;</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="q">&quot;, &quot;</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
1166 1153 <span class="s">}</span>
1167 1154
1168 1155 <span class="i">$SpecifiedComparisonMeasureMethod</span> = <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
1169 1156 <span class="i">$SpecifiedComparisonMeasureName</span> = <span class="i">$SupportedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
1170 1157
1171 1158 <span class="i">$OptionsInfo</span>{<span class="w">BitVectorComparisonMode</span>} = <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>}<span class="sc">;</span>
1172 1159
1173 1160 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonMeasure</span>} = <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
1174 1161 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonMeasureName</span>} = <span class="i">$SpecifiedComparisonMeasureName</span><span class="sc">;</span>
1175 1162 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonMeasureMethod</span>} = <span class="i">$SpecifiedComparisonMeasureMethod</span><span class="sc">;</span>
1176 1163
1177 1164 <span class="c"># Make sure valid alpha parameter is specified for Tversky calculation...</span>
1178 1165 <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1179 1166 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedMeasure</span> =~ <span class="q">/^(TverskySimilarity|WeightedTverskySimilarity)$/i</span><span class="s">)</span> <span class="s">{</span>
1180 1167 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">alpha</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1181 1168 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;-a, --alpha\&quot; option in \&quot;TverskySimilarity or WeightedTverskySimilarity\&quot; \&quot;-m --mode\&quot;. \n&quot;</span><span class="sc">;</span>
1182 1169 <span class="s">}</span>
1183 1170 <span class="k">my</span><span class="s">(</span><span class="i">$Alpha</span><span class="s">)</span><span class="sc">;</span>
1184 1171 <span class="i">$Alpha</span> = <span class="i">$Options</span>{<span class="w">alpha</span>}<span class="sc">;</span>
1185 1172 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsFloat</span><span class="s">(</span><span class="i">$Alpha</span><span class="s">)</span> &amp;&amp; <span class="i">$Alpha</span> &gt;=<span class="n">0</span> &amp;&amp; <span class="i">$Alpha</span> &lt;= <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1186 1173 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{alpha}, for option \&quot;-a, --alpha\&quot; is not valid. Allowed values: &gt;= 0 and &lt;= 1\n&quot;</span><span class="sc">;</span>
1187 1174 <span class="s">}</span>
1188 1175 <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>} = <span class="i">$Alpha</span><span class="sc">;</span>
1189 1176 <span class="s">}</span>
1190 1177
1191 1178 <span class="c"># Make sure valid beta parameter is specified for WeightedTanimoto and WeightedTversky</span>
1192 1179 <span class="c"># calculations...</span>
1193 1180 <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1194 1181 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedMeasure</span> =~ <span class="q">/^(WeightedTverskySimilarity|WeightedTanimotoSimilarity)$/i</span><span class="s">)</span> <span class="s">{</span>
1195 1182 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">beta</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1196 1183 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;-b, --beta\&quot; option in \&quot;WeightedTverskySimilarity or WeightedTanimotoSimilarity\&quot; \&quot;-m --mode\&quot;. \n&quot;</span><span class="sc">;</span>
1197 1184 <span class="s">}</span>
1198 1185 <span class="k">my</span><span class="s">(</span><span class="i">$Beta</span><span class="s">)</span><span class="sc">;</span>
1199 1186 <span class="i">$Beta</span> = <span class="i">$Options</span>{<span class="w">beta</span>}<span class="sc">;</span>
1200 1187 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsFloat</span><span class="s">(</span><span class="i">$Beta</span><span class="s">)</span> &amp;&amp; <span class="i">$Beta</span> &gt;=<span class="n">0</span> &amp;&amp; <span class="i">$Beta</span> &lt;= <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1201 1188 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{beta}, for option \&quot;-b, --beta\&quot; is not valid. Allowed values: &gt;= 0 and &lt;= 1\n&quot;</span><span class="sc">;</span>
1202 1189 <span class="s">}</span>
1203 1190 <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>} = <span class="i">$Beta</span><span class="sc">;</span>
1204 1191 <span class="s">}</span>
1205 1192 <span class="s">}</span>
1206 1193
1207 1194 <span class="c"># Process options related to comparion of vector strings...</span>
1208 1195 <span class="c">#</span>
1209 <a name="ProcessVectorComparisonOptions-"></a>1196 <span class="k">sub </span><span class="m">ProcessVectorComparisonOptions</span> <span class="s">{</span>
1210 1197 <span class="c"># Setup specified similarity coefficients for vector strings..</span>
1211 1198 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$SupportedComparisonMeasure</span><span class="cm">,</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresNameMap</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresMethodMap</span><span class="s">)</span><span class="sc">;</span>
1212 1199
1213 1200 <span class="i">@SupportedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1214 1201 <span class="i">%SupportedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1215 1202 <span class="i">%SupportedComparisonMeasuresMethodMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1216 1203 <span class="k">for</span> <span class="i">$SupportedComparisonMeasure</span> <span class="s">(</span><span class="i">Fingerprints::FingerprintsVector::GetSupportedDistanceAndSimilarityCoefficients</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1217 1204 <span class="c"># Similarity and distance coefficient function/method names contain &quot;Coefficient&quot; in their names.</span>
1218 1205 <span class="c"># So take &#39;em out and setup a map to original function/method name...</span>
1219 1206 <span class="i">$ComparisonMeasure</span> = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
1220 1207 <span class="k">if</span> <span class="s">(</span><span class="i">$ComparisonMeasure</span> =~ <span class="q">/Coefficient$/i</span><span class="s">)</span> <span class="s">{</span>
1221 1208 <span class="i">$ComparisonMeasure</span> =~ <span class="q">s/Coefficient$//i</span><span class="sc">;</span>
1222 1209 <span class="s">}</span>
1223 1210 <span class="k">push</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
1224 1211 <span class="i">$SupportedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>} = <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
1225 1212 <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>} = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
1226 1213 <span class="s">}</span>
1227 1214
1228 1215 <span class="c"># Setup a list of similarity coefficients to use for calculating similarity matrices for bit vector strings...</span>
1229 1216 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="cm">,</span> <span class="i">$SpecifiedComparisonMeasureName</span><span class="cm">,</span> <span class="i">$SpecifiedComparisonMeasureMethod</span><span class="s">)</span><span class="sc">;</span>
1230 1217
1231 1218 <span class="i">$SpecifiedComparisonMeasureName</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1232 1219 <span class="i">$SpecifiedComparisonMeasureMethod</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1233 1220
1234 1221 <span class="i">$SpecifiedMeasure</span> = <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>}<span class="sc">;</span>
1235 1222 <span class="i">$SpecifiedMeasure</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
1236 1223
1237 1224 <span class="k">if</span> <span class="s">(</span>! <span class="k">exists</span><span class="s">(</span><span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1238 1225 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $SpecifiedMeasure, for option \&quot;-v --VectorComparisonMode\&quot; is not valid.\nAllowed values:&quot;</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="q">&quot;, &quot;</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
1239 1226 <span class="s">}</span>
1240 1227
1241 1228 <span class="i">$SpecifiedComparisonMeasureMethod</span> = <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
1242 1229 <span class="i">$SpecifiedComparisonMeasureName</span> = <span class="i">$SupportedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
1243 1230
1244 1231 <span class="i">$OptionsInfo</span>{<span class="w">VectorComparisonMode</span>} = <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>}<span class="sc">;</span>
1245 1232
1246 1233 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMeasure</span>} = <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
1247 1234 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMeasuresName</span>} = <span class="i">$SpecifiedComparisonMeasureName</span><span class="sc">;</span>
1248 1235 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMeasuresMethod</span>} = <span class="i">$SpecifiedComparisonMeasureMethod</span><span class="sc">;</span>
1249 1236
1250 1237 <span class="c"># Setup specified vector comparison calculation modes...</span>
1251 1238 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedFormulism</span><span class="s">)</span><span class="sc">;</span>
1252 1239
1253 1240 <span class="i">$SpecifiedFormulism</span> = <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>}<span class="sc">;</span>
1254 1241 <span class="i">$SpecifiedFormulism</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
1255 1242 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedFormulism</span> !~ <span class="q">/^(AlgebraicForm|BinaryForm|SetTheoreticForm)$/i</span><span class="s">)</span> <span class="s">{</span>
1256 1243 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $SpecifiedFormulism, for option \&quot;--VectorComparisonFormulism\&quot; is not valid. Allowed values: AlgebraicForm, BinaryForm or SetTheoreticForm\n&quot;</span><span class="sc">;</span>
1257 1244 <span class="s">}</span>
1258 1245
1259 1246 <span class="i">$OptionsInfo</span>{<span class="w">VectorComparisonFormulism</span>} = <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>}<span class="sc">;</span>
1260 1247 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonMode</span>} = <span class="i">$SpecifiedFormulism</span><span class="sc">;</span>
1261 1248
1262 1249 <span class="s">}</span>
1263 1250
1264 1251 <span class="c"># Process options related to data retrieval from reference fingerprints SD and CSV/TSV</span>
1265 1252 <span class="c"># text files...</span>
1266 1253 <span class="c">#</span>
1267 <a name="ProcessReferenceFingerprintsDataOptions-"></a>1254 <span class="k">sub </span><span class="m">ProcessReferenceFingerprintsDataOptions</span> <span class="s">{</span>
1268 1255
1269 1256 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDPrefix</span>} = <span class="i">$Options</span>{<span class="w">referencecompoundidprefix</span>} ? <span class="i">$Options</span>{<span class="w">referencecompoundidprefix</span>} <span class="co">:</span> <span class="q">&#39;Cmpd&#39;</span><span class="sc">;</span>
1270 1257
1271 1258 <span class="c"># Compound ID and fingerprints column options for text files...</span>
1272 1259
1273 1260 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceColMode</span>} = <span class="i">$Options</span>{<span class="w">referencecolmode</span>}<span class="sc">;</span>
1274 1261
1275 1262 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1276 1263 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecolmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
1277 1264 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1278 1265 <span class="k">die</span> <span class="q">&quot;Error: Column value, $Options{referencecompoundidcol}, specified using \&quot;--ReferenceCompoundIDCol\&quot; is not valid: Allowed integer values: &gt; 0\n&quot;</span><span class="sc">;</span>
1279 1266 <span class="s">}</span>
1280 1267 <span class="s">}</span>
1281 1268 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDCol</span>} = <span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>}<span class="sc">;</span>
1282 1269 <span class="s">}</span>
1283 1270 <span class="k">else</span> <span class="s">{</span>
1284 1271 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDCol</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1285 1272 <span class="s">}</span>
1286 1273
1287 1274 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1288 1275 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecolmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
1289 1276 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1290 1277 <span class="k">die</span> <span class="q">&quot;Error: Column value, $Options{referencefingerprintscol}, specified using \&quot;--ReferenceFingerprintsCol\&quot; is not valid: Allowed integer values: &gt; 0\n&quot;</span><span class="sc">;</span>
1291 1278 <span class="s">}</span>
1292 1279 <span class="s">}</span>
1293 1280 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceFingerprintsCol</span>} = <span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="sc">;</span>
1294 1281 <span class="s">}</span>
1295 1282 <span class="k">else</span> <span class="s">{</span>
1296 1283 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceFingerprintsCol</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1297 1284 <span class="s">}</span>
1298 1285
1299 1286 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>}<span class="s">)</span> &amp;&amp; <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1300 1287 <span class="k">if</span> <span class="s">(</span><span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>}<span class="s">)</span> &amp;&amp; <span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1301 1288 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>} == <span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1302 1289 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--ReferenceCompoundIDCol\&quot; and \&quot;--ReferenceFingerprintsCol\&quot;, $Options{referencecompoundidcol}, must be different.\n&quot;</span><span class="sc">;</span>
1303 1290 <span class="s">}</span>
1304 1291 <span class="s">}</span>
1305 1292 <span class="k">else</span> <span class="s">{</span>
1306 1293 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidcol</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">referencefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1307 1294 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--ReferenceCompoundIDCol\&quot; and \&quot;--ReferenceFingerprintsCol\&quot;, $Options{referencecompoundidcol}, must be different.\n&quot;</span><span class="sc">;</span>
1308 1295 <span class="s">}</span>
1309 1296 <span class="s">}</span>
1310 1297 <span class="s">}</span>
1311 1298
1312 1299 <span class="c"># Compound ID and fingerprints field options for SD files...</span>
1313 1300
1314 1301 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDMode</span>} = <span class="i">$Options</span>{<span class="w">referencecompoundidmode</span>}<span class="sc">;</span>
1315 1302 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDField</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1316 1303
1317 1304 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidmode</span>} =~ <span class="q">/^DataField$/i</span> &amp;&amp; !<span class="i">$Options</span>{<span class="w">referencecompoundidfield</span>}<span class="s">)</span> <span class="s">{</span>
1318 1305 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--ReferenceCompoundIDField\&quot; option in \&quot;DataField\&quot; \&quot;--ReferenceCompoundIDMode\&quot;. \n&quot;</span><span class="sc">;</span>
1319 1306 <span class="s">}</span>
1320 1307 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidfield</span>}<span class="s">)</span> <span class="s">{</span>
1321 1308 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceCompoundIDField</span>} = <span class="i">$Options</span>{<span class="w">referencecompoundidfield</span>}<span class="sc">;</span>
1322 1309 <span class="s">}</span>
1323 1310
1324 1311 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1325 1312 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceFingerprintsField</span>} = <span class="i">$Options</span>{<span class="w">referencefingerprintsfield</span>}<span class="sc">;</span>
1326 1313 <span class="s">}</span>
1327 1314 <span class="k">else</span> <span class="s">{</span>
1328 1315 <span class="i">$OptionsInfo</span>{<span class="w">ReferenceFingerprintsField</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1329 1316 <span class="s">}</span>
1330 1317
1331 1318 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidfield</span>} &amp;&amp; <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1332 1319 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidfield</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">referencefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1333 1320 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--ReferenceCompoundIDField\&quot; and \&quot;--ReferenceFingerprintsfield\&quot;, $Options{referencecompoundidfield}, must be different.\n&quot;</span><span class="sc">;</span>
1334 1321 <span class="s">}</span>
1335 1322 <span class="s">}</span>
1336 1323
1337 1324 <span class="s">}</span>
1338 1325
1339 1326 <span class="c"># Process options related to data retrieval from database fingerprints SD and CSV/TSV</span>
1340 1327 <span class="c"># text files...</span>
1341 1328 <span class="c">#</span>
1342 <a name="ProcessDatabaseFingerprintsDataOptions-"></a>1329 <span class="k">sub </span><span class="m">ProcessDatabaseFingerprintsDataOptions</span> <span class="s">{</span>
1343 1330
1344 1331 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDPrefix</span>} = <span class="i">$Options</span>{<span class="w">databasecompoundidprefix</span>} ? <span class="i">$Options</span>{<span class="w">databasecompoundidprefix</span>} <span class="co">:</span> <span class="q">&#39;Cmpd&#39;</span><span class="sc">;</span>
1345 1332
1346 1333 <span class="c"># Compound ID and fingerprints column options for text files...</span>
1347 1334
1348 1335 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseColMode</span>} = <span class="i">$Options</span>{<span class="w">databasecolmode</span>}<span class="sc">;</span>
1349 1336
1350 1337 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1351 1338 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecolmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
1352 1339 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1353 1340 <span class="k">die</span> <span class="q">&quot;Error: Column value, $Options{databasecompoundidcol}, specified using \&quot;--DatabaseCompoundIDCol\&quot; is not valid: Allowed integer values: &gt; 0\n&quot;</span><span class="sc">;</span>
1354 1341 <span class="s">}</span>
1355 1342 <span class="s">}</span>
1356 1343 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDCol</span>} = <span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>}<span class="sc">;</span>
1357 1344 <span class="s">}</span>
1358 1345 <span class="k">else</span> <span class="s">{</span>
1359 1346 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDCol</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1360 1347 <span class="s">}</span>
1361 1348
1362 1349 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1363 1350 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecolmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
1364 1351 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1365 1352 <span class="k">die</span> <span class="q">&quot;Error: Column value, $Options{databasefingerprintscol}, specified using \&quot;--DatabaseFingerprintsCol\&quot; is not valid: Allowed integer values: &gt; 0\n&quot;</span><span class="sc">;</span>
1366 1353 <span class="s">}</span>
1367 1354 <span class="s">}</span>
1368 1355 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseFingerprintsCol</span>} = <span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="sc">;</span>
1369 1356 <span class="s">}</span>
1370 1357 <span class="k">else</span> <span class="s">{</span>
1371 1358 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseFingerprintsCol</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1372 1359 <span class="s">}</span>
1373 1360
1374 1361 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>}<span class="s">)</span> &amp;&amp; <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1375 1362 <span class="k">if</span> <span class="s">(</span><span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>}<span class="s">)</span> &amp;&amp; <span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1376 1363 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>} == <span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1377 1364 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--DatabaseCompoundIDCol\&quot; and \&quot;--DatabaseFingerprintsCol\&quot;, $Options{databasecompoundidcol}, must be different.\n&quot;</span><span class="sc">;</span>
1378 1365 <span class="s">}</span>
1379 1366 <span class="s">}</span>
1380 1367 <span class="k">else</span> <span class="s">{</span>
1381 1368 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidcol</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">databasefingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1382 1369 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--DatabaseCompoundIDCol\&quot; and \&quot;--DatabaseFingerprintsCol\&quot;, $Options{databasecompoundidcol}, must be different.\n&quot;</span><span class="sc">;</span>
1383 1370 <span class="s">}</span>
1384 1371 <span class="s">}</span>
1385 1372 <span class="s">}</span>
1386 1373
1387 1374 <span class="c"># Database data column options for text files...</span>
1388 1375
1389 1376 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataColsMode</span>} = <span class="i">$Options</span>{<span class="w">databasedatacolsmode</span>}<span class="sc">;</span>
1390 1377 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataCols</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1391 1378 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataCols</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1392 1379
1393 1380 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatacolsmode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
1394 1381 <span class="k">my</span><span class="s">(</span><span class="i">$DatabaseDataCols</span><span class="cm">,</span> <span class="i">$DatabaseColNum</span><span class="cm">,</span> <span class="i">@SpecifiedDataCols</span><span class="s">)</span><span class="sc">;</span>
1395 1382
1396 1383 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">databasedatacols</span>}<span class="s">)</span> <span class="s">{</span>
1397 1384 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--DatabaseDataCols\&quot; option in \&quot;Specify\&quot; \&quot;--DatabaseDataColsMode\&quot;. \n&quot;</span><span class="sc">;</span>
1398 1385 <span class="s">}</span>
1399 1386 <span class="i">$DatabaseDataCols</span> = <span class="i">$Options</span>{<span class="w">databasedatacols</span>}<span class="sc">;</span>
1400 1387
1401 1388 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecolmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
1402 1389 <span class="i">$DatabaseDataCols</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
1403 1390 <span class="i">@SpecifiedDataCols</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$DatabaseDataCols</span><span class="sc">;</span>
1404 1391 <span class="k">for</span> <span class="i">$DatabaseColNum</span> <span class="s">(</span><span class="i">@SpecifiedDataCols</span><span class="s">)</span> <span class="s">{</span>
1405 1392 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$DatabaseColNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1406 1393 <span class="k">die</span> <span class="q">&quot;Error: Column value, $DatabaseColNum, specified using \&quot;--DatabaseDataCols\&quot; is not valid: Allowed integer values: &gt; 0\n&quot;</span><span class="sc">;</span>
1407 1394 <span class="s">}</span>
1408 1395 <span class="s">}</span>
1409 1396 <span class="s">}</span>
1410 1397 <span class="k">else</span> <span class="s">{</span>
1411 1398 <span class="i">@SpecifiedDataCols</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$DatabaseDataCols</span><span class="sc">;</span>
1412 1399 <span class="s">}</span>
1413 1400 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataCols</span>} = <span class="i">$DatabaseDataCols</span><span class="sc">;</span>
1414 1401 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataCols</span>}}<span class="cm">,</span> <span class="i">@SpecifiedDataCols</span><span class="sc">;</span>
1415 1402 <span class="s">}</span>
1416 1403 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatacolsmode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
1417 1404 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataCols</span>} = <span class="q">&#39;All&#39;</span><span class="sc">;</span>
1418 1405 <span class="s">}</span>
1419 1406
1420 1407 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataColsMode</span>} =~ <span class="q">/^Specify$/i</span> &amp;&amp; !<span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataCols</span>}<span class="s">)</span> <span class="s">{</span>
1421 1408 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--DatabaseDataCols\&quot; option in \&quot;Specify\&quot; \&quot;--DatabaseDataColsMode\&quot;. \n&quot;</span><span class="sc">;</span>
1422 1409 <span class="s">}</span>
1423 1410
1424 1411 <span class="c"># Compound ID and fingerprints field options for SD files...</span>
1425 1412
1426 1413 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDMode</span>} = <span class="i">$Options</span>{<span class="w">databasecompoundidmode</span>}<span class="sc">;</span>
1427 1414 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDField</span>} = <span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>} ? <span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="sc">;</span>
1428 1415
1429 1416 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidmode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span>
1430 1417 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>}<span class="s">)</span> <span class="s">{</span>
1431 1418 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--DatabaseCompoundIDField\&quot; option in \&quot;DataField\&quot; \&quot;--DatabaseCompoundIDMode\&quot;. \n&quot;</span><span class="sc">;</span>
1432 1419 <span class="s">}</span>
1433 1420 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseCompoundIDField</span>} = <span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>}<span class="sc">;</span>
1434 1421 <span class="s">}</span>
1435 1422
1436 1423
1437 1424 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1438 1425 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseFingerprintsField</span>} = <span class="i">$Options</span>{<span class="w">databasefingerprintsfield</span>}<span class="sc">;</span>
1439 1426 <span class="s">}</span>
1440 1427 <span class="k">else</span> <span class="s">{</span>
1441 1428 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseFingerprintsField</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1442 1429 <span class="s">}</span>
1443 1430
1444 1431 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>} &amp;&amp; <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1445 1432 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidfield</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">databasefingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1446 1433 <span class="k">die</span> <span class="q">&quot;Error: Values specified using \&quot;--DatabaseCompoundIDField\&quot; and \&quot;--DatabaseFingerprintsfield\&quot;, $Options{databasecompoundidfield}, must be different.\n&quot;</span><span class="sc">;</span>
1447 1434 <span class="s">}</span>
1448 1435 <span class="s">}</span>
1449 1436
1450 1437 <span class="c"># Database data field options for SD files...</span>
1451 1438
1452 1439 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFieldsMode</span>} = <span class="i">$Options</span>{<span class="w">databasedatafieldsmode</span>}<span class="sc">;</span>
1453 1440 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFields</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
1454 1441 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1455 1442
1456 1443 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatafieldsmode</span>} =~ <span class="q">/^Specify$/i</span> &amp;&amp; !<span class="i">$Options</span>{<span class="w">databasedatafields</span>}<span class="s">)</span> <span class="s">{</span>
1457 1444 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--DatabaseDataFields\&quot; option in \&quot;Specify\&quot; \&quot;--DatabaseDataFieldsMode\&quot;. \n&quot;</span><span class="sc">;</span>
1458 1445 <span class="s">}</span>
1459 1446 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatafields</span>}<span class="s">)</span> <span class="s">{</span>
1460 1447 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedDataFields</span><span class="s">)</span><span class="sc">;</span>
1461 1448 <span class="i">$OptionsInfo</span>{<span class="w">DatabaseDataFields</span>} = <span class="i">$Options</span>{<span class="w">databasedatafields</span>}<span class="sc">;</span>
1462 1449
1463 1450 <span class="i">@SpecifiedDataFields</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">databasedatafields</span>}<span class="sc">;</span>
1464 1451 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDatabaseDataFields</span>}}<span class="cm">,</span> <span class="i">@SpecifiedDataFields</span><span class="sc">;</span>
1465 1452 <span class="s">}</span>
1466 1453 <span class="s">}</span>
1467 1454
1468 1455 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
1469 <a name="SetupScriptUsage-"></a>1456 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
1470 1457
1471 1458 <span class="c"># Retrieve all the options...</span>
1472 1459 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
1473 1460
1474 1461 <span class="i">$Options</span>{<span class="w">alpha</span>} = <span class="n">0.5</span><span class="sc">;</span>
1475 1462 <span class="i">$Options</span>{<span class="w">beta</span>} = <span class="n">1</span><span class="sc">;</span>
1476 1463
1477 1464 <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>} = <span class="q">&quot;TanimotoSimilarity&quot;</span><span class="sc">;</span>
1478 1465
1479 1466 <span class="i">$Options</span>{<span class="w">databasecolmode</span>} = <span class="q">&#39;colnum&#39;</span><span class="sc">;</span>
1480 1467
1481 1468 <span class="i">$Options</span>{<span class="w">databasecompoundidprefix</span>} = <span class="q">&#39;Cmpd&#39;</span><span class="sc">;</span>
1482 1469 <span class="i">$Options</span>{<span class="w">databasecompoundidmode</span>} = <span class="q">&#39;LabelPrefix&#39;</span><span class="sc">;</span>
1483 1470
1484 1471 <span class="i">$Options</span>{<span class="w">databasedatacolsmode</span>} = <span class="q">&#39;CompoundID&#39;</span><span class="sc">;</span>
1485 1472 <span class="i">$Options</span>{<span class="w">databasedatafieldsmode</span>} = <span class="q">&#39;CompoundID&#39;</span><span class="sc">;</span>
1486 1473
1487 1474 <span class="i">$Options</span>{<span class="w">distancecutoff</span>} = <span class="n">10</span><span class="sc">;</span>
1488 1475
1489 1476 <span class="i">$Options</span>{<span class="w">referencecolmode</span>} = <span class="q">&#39;colnum&#39;</span><span class="sc">;</span>
1490 1477
1491 1478 <span class="i">$Options</span>{<span class="w">referencecompoundidprefix</span>} = <span class="q">&#39;Cmpd&#39;</span><span class="sc">;</span>
1492 1479 <span class="i">$Options</span>{<span class="w">referencecompoundidmode</span>} = <span class="q">&#39;LabelPrefix&#39;</span><span class="sc">;</span>
1493 1480
1494 1481 <span class="i">$Options</span>{<span class="w">detail</span>} = <span class="n">1</span><span class="sc">;</span>
1495 1482
1496 1483 <span class="i">$Options</span>{<span class="w">fingerprintsmode</span>} = <span class="q">&#39;AutoDetect&#39;</span><span class="sc">;</span>
1497 1484 <span class="i">$Options</span>{<span class="w">groupfusionrule</span>} = <span class="q">&#39;Max&#39;</span><span class="sc">;</span>
1498 1485 <span class="i">$Options</span>{<span class="w">groupfusionapplycutoff</span>} = <span class="q">&#39;Yes&#39;</span><span class="sc">;</span>
1499 1486
1500 1487 <span class="i">$Options</span>{<span class="w">knn</span>} = <span class="q">&#39;All&#39;</span><span class="sc">;</span>
1501 1488
1502 1489 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">&#39;MultipleReferences&#39;</span><span class="sc">;</span>
1503 1490
1504 1491 <span class="i">$Options</span>{<span class="w">numofsimilarmolecules</span>} = <span class="n">10</span><span class="sc">;</span>
1505 1492 <span class="i">$Options</span>{<span class="w">percentsimilarmolecules</span>} = <span class="n">1</span><span class="sc">;</span>
1506 1493
1507 1494 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">&#39;comma&#39;</span><span class="sc">;</span>
1508 1495 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">&#39;comma&#39;</span><span class="sc">;</span>
1509 1496 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">&#39;yes&#39;</span><span class="sc">;</span>
1510 1497
1511 1498 <span class="i">$Options</span>{<span class="w">output</span>} = <span class="q">&#39;text&#39;</span><span class="sc">;</span>
1512 1499
1513 1500 <span class="i">$Options</span>{<span class="w">precision</span>} = <span class="n">2</span><span class="sc">;</span>
1514 1501
1515 1502 <span class="i">$Options</span>{<span class="w">searchmode</span>} = <span class="q">&#39;SimilaritySearch&#39;</span><span class="sc">;</span>
1516 1503
1517 1504 <span class="i">$Options</span>{<span class="w">similarcountmode</span>} = <span class="q">&#39;NumOfSimilar&#39;</span><span class="sc">;</span>
1518 1505
1519 1506 <span class="i">$Options</span>{<span class="w">similaritycutoff</span>} = <span class="n">0.75</span><span class="sc">;</span>
1520 1507
1521 1508 <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>} = <span class="q">&#39;TanimotoSimilarity&#39;</span><span class="sc">;</span>
1522 1509 <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>} = <span class="q">&#39;AlgebraicForm&#39;</span><span class="sc">;</span>
1523 1510
1524 1511 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">&quot;alpha=f&quot;</span><span class="cm">,</span> <span class="q">&quot;beta=f&quot;</span><span class="cm">,</span> <span class="q">&quot;bitvectorcomparisonmode|b=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasecolmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasecompoundidcol=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasecompoundidprefix=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasecompoundidfield=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasecompoundidmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasedatacols=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasedatacolsmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasedatafields=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasedatafieldsmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasefingerprintscol=s&quot;</span><span class="cm">,</span> <span class="q">&quot;databasefingerprintsfield=s&quot;</span><span class="cm">,</span> <span class="q">&quot;distancecutoff=f&quot;</span><span class="cm">,</span> <span class="q">&quot;detail|d=i&quot;</span><span class="cm">,</span> <span class="q">&quot;fast|f&quot;</span><span class="cm">,</span> <span class="q">&quot;fingerprintsmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;groupfusionrule|g=s&quot;</span><span class="cm">,</span> <span class="cm">,</span> <span class="q">&quot;groupfusionapplycutoff=s&quot;</span><span class="cm">,</span> <span class="q">&quot;help|h&quot;</span><span class="cm">,</span> <span class="q">&quot;indelim=s&quot;</span><span class="cm">,</span> <span class="q">&quot;knn|k=s&quot;</span><span class="cm">,</span> <span class="q">&quot;mode|m=s&quot;</span><span class="cm">,</span> <span class="q">&quot;numofsimilarmolecules|n=i&quot;</span><span class="cm">,</span> <span class="q">&quot;outdelim=s&quot;</span><span class="cm">,</span> <span class="q">&quot;output=s&quot;</span><span class="cm">,</span> <span class="q">&quot;overwrite|o&quot;</span><span class="cm">,</span> <span class="q">&quot;percentsimilarmolecules|p=f&quot;</span><span class="cm">,</span> <span class="q">&quot;precision=s&quot;</span><span class="cm">,</span> <span class="q">&quot;quote|q=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencecolmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencecompoundidcol=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencecompoundidprefix=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencecompoundidfield=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencecompoundidmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencefingerprintscol=s&quot;</span><span class="cm">,</span> <span class="q">&quot;referencefingerprintsfield=s&quot;</span><span class="cm">,</span> <span class="q">&quot;root|r=s&quot;</span><span class="cm">,</span> <span class="q">&quot;searchmode|s=s&quot;</span><span class="cm">,</span> <span class="q">&quot;similarcountmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;similaritycutoff=f&quot;</span><span class="cm">,</span> <span class="q">&quot;vectorcomparisonmode|v=s&quot;</span><span class="cm">,</span> <span class="q">&quot;vectorcomparisonformulism=s&quot;</span><span class="cm">,</span> <span class="q">&quot;workingdir|w=s&quot;</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1525 1512 <span class="k">die</span> <span class="q">&quot;\nTo get a list of valid options and their values, use \&quot;$ScriptName -h\&quot; or\n\&quot;perl -S $ScriptName -h\&quot; command and try again...\n&quot;</span><span class="sc">;</span>
1526 1513 <span class="s">}</span>
1527 1514 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
1528 1515 <span class="k">if</span> <span class="s">(</span>! <span class="k">-d</span> <span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
1529 1516 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{workingdir}, for option \&quot;-w --workingdir\&quot; is not a directory name.\n&quot;</span><span class="sc">;</span>
1530 1517 <span class="s">}</span>
1531 1518 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t chdir $Options{workingdir}: $! \n&quot;</span><span class="sc">;</span>
1532 1519 <span class="s">}</span>
1533 1520 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecolmode</span>} !~ <span class="q">/^(ColNum|ColLabel)$/i</span><span class="s">)</span> <span class="s">{</span>
1534 1521 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{databasecolmode}, for option \&quot;--DatabaseColMode\&quot; is not valid. Allowed values: ColNum, or ColLabel\n&quot;</span><span class="sc">;</span>
1535 1522 <span class="s">}</span>
1536 1523 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasecompoundidmode</span>} !~ <span class="q">/^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
1537 1524 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{databasecompoundidmode}, for option \&quot;--DatabaseCompoundIDMode\&quot; is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n&quot;</span><span class="sc">;</span>
1538 1525 <span class="s">}</span>
1539 1526 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatacolsmode</span>} !~ <span class="q">/^(All|Specify|CompoundID)$/i</span><span class="s">)</span> <span class="s">{</span>
1540 1527 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{databasedatacolsmode}, for option \&quot;--DatabaseDataColsMode\&quot; is not valid. Allowed values: All, Specify, or CompoundID\n&quot;</span><span class="sc">;</span>
1541 1528 <span class="s">}</span>
1542 1529 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">databasedatafieldsmode</span>} !~ <span class="q">/^(All|Common|Specify|CompoundID)$/i</span><span class="s">)</span> <span class="s">{</span>
1543 1530 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{databasedatafieldsmode}, for option \&quot;--DatabaseDataFieldsMode\&quot; is not valid. Allowed values: All, Common, Specify, or CompoundID\n&quot;</span><span class="sc">;</span>
1544 1531 <span class="s">}</span>
1545 1532 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">detail</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1546 1533 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{detail}, for option \&quot;-d, --detail\&quot; is not valid. Allowed values: &gt; 0 \n&quot;</span><span class="sc">;</span>
1547 1534 <span class="s">}</span>
1548 1535 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintsmode</span>} !~ <span class="q">/^(AutoDetect|FingerprintsBitVectorString|FingerprintsVectorString)$/i</span><span class="s">)</span> <span class="s">{</span>
1549 1536 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{fingerprintsmode}, for option \&quot;--FingerprintsMode\&quot; is not valid. Allowed values: AutoDetect, FingerprintsBitVectorString or FingerprintsVectorString \n&quot;</span><span class="sc">;</span>
1550 1537 <span class="s">}</span>
1551 1538 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">groupfusionrule</span>} !~ <span class="q">/^(Max|Min|Mean|Median|Sum|Euclidean)$/i</span><span class="s">)</span> <span class="s">{</span>
1552 1539 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{groupfusionrule}, for option \&quot;-g, --GroupFusionRule\&quot; is not valid. Allowed values: Max, Min, Mean, Median, Sum, Euclidean\n&quot;</span><span class="sc">;</span>
1553 1540 <span class="s">}</span>
1554 1541 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">groupfusionapplycutoff</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
1555 1542 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{quote}, for option \&quot;--GroupFusionApplyCutoff\&quot; is not valid. Allowed values: Yes or No\n&quot;</span><span class="sc">;</span>
1556 1543 <span class="s">}</span>
1557 1544 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span>
1558 1545 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{indelim}, for option \&quot;--InDelim\&quot; is not valid. Allowed values: comma, or semicolon\n&quot;</span><span class="sc">;</span>
1559 1546 <span class="s">}</span>
1560 1547 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(IndividualReference|MultipleReferences)$/i</span><span class="s">)</span> <span class="s">{</span>
1561 1548 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{mode}, for option \&quot;-m, --mode\&quot; is not valid. Allowed values: IndividualReference, MultipleReferences\n&quot;</span><span class="sc">;</span>
1562 1549 <span class="s">}</span>
1563 1550 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">numofsimilarmolecules</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1564 1551 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{numofsimilarmolecules}, for option \&quot;-n, --NumOfSimilarMolecules\&quot; is not valid. Allowed values: &gt; 0 \n&quot;</span><span class="sc">;</span>
1565 1552 <span class="s">}</span>
1566 1553 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
1567 1554 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{outdelim}, for option \&quot;--OutDelim\&quot; is not valid. Allowed values: comma, tab, or semicolon\n&quot;</span><span class="sc">;</span>
1568 1555 <span class="s">}</span>
1569 1556 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} !~ <span class="q">/^(SD|text|both)$/i</span><span class="s">)</span> <span class="s">{</span>
1570 1557 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{output}, for option \&quot;--output\&quot; is not valid. Allowed values: SD, text, or both\n&quot;</span><span class="sc">;</span>
1571 1558 <span class="s">}</span>
1572 1559 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsFloat</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">percentsimilarmolecules</span>}<span class="s">)</span> &amp;&amp; <span class="i">$Options</span>{<span class="w">percentsimilarmolecules</span>} &gt; <span class="n">0</span> &amp;&amp; <span class="i">$Options</span>{<span class="w">percentsimilarmolecules</span>} &lt;= <span class="n">100</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1573 1560 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{percentsimilarmolecules}, for option \&quot;-p, --PercentSimilarMolecules\&quot; is not valid. Allowed values: &gt; 0 and &lt;= 100 \n&quot;</span><span class="sc">;</span>
1574 1561 <span class="s">}</span>
1575 1562 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
1576 1563 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{quote}, for option \&quot;-q --quote\&quot; is not valid. Allowed values: Yes or No\n&quot;</span><span class="sc">;</span>
1577 1564 <span class="s">}</span>
1578 1565 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">precision</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
1579 1566 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{precision}, for option \&quot;--precision\&quot; is not valid. Allowed values: &gt; 0 \n&quot;</span><span class="sc">;</span>
1580 1567 <span class="s">}</span>
1581 1568 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecolmode</span>} !~ <span class="q">/^(ColNum|ColLabel)$/i</span><span class="s">)</span> <span class="s">{</span>
1582 1569 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{referencecolmode}, for option \&quot;--ReferenceColMode\&quot; is not valid. Allowed values: ColNum, or ColLabel\n&quot;</span><span class="sc">;</span>
1583 1570 <span class="s">}</span>
1584 1571 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">referencecompoundidmode</span>} !~ <span class="q">/^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
1585 1572 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{referencecompoundidmode}, for option \&quot;--ReferenceCompoundIDMode\&quot; is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n&quot;</span><span class="sc">;</span>
1586 1573 <span class="s">}</span>
1587 1574 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">searchmode</span>} !~ <span class="q">/^(SimilaritySearch|DissimilaritySearch)$/i</span><span class="s">)</span> <span class="s">{</span>
1588 1575 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{searchmode}, for option \&quot;-s, --SearchMode\&quot; is not valid. Allowed values: SimilaritySearch, DissimilaritySearch \n&quot;</span><span class="sc">;</span>
1589 1576 <span class="s">}</span>
1590 1577 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">similarcountmode</span>} !~ <span class="q">/^(NumOfSimilar|PercentSimilar)$/i</span><span class="s">)</span> <span class="s">{</span>
1591 1578 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{similarcountmode}, for option \&quot;--SimilarCountMode\&quot; is not valid. Allowed values: NumOfSimilar, PercentSimilar \n&quot;</span><span class="sc">;</span>
1592 1579 <span class="s">}</span>
1593 1580 <span class="s">}</span>
1594 1581
1595 <a name="EOF-"></a></pre>
1596 <p>&nbsp;</p>
1597 <br />
1598 <center>
1599 <img src="../../../images/h2o2.png">
1600 </center>
1601 </body>
1602 </html>