|
0
|
1 <html>
|
|
|
2 <head>
|
|
|
3 <title>MayaChemTools:Code:SimilarityMatricesFingerprints.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: SimilarityMatricesFingerprints.pl,v $</span>
|
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:20 $</span>
|
|
|
18 5 <span class="c"># $Revision: 1.21 $</span>
|
|
|
19 6 <span class="c">#</span>
|
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
|
21 8 <span class="c">#</span>
|
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
|
23 10 <span class="c">#</span>
|
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
|
25 12 <span class="c">#</span>
|
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
|
29 16 <span class="c"># later version.</span>
|
|
|
30 17 <span class="c">#</span>
|
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
|
34 21 <span class="c"># details.</span>
|
|
|
35 22 <span class="c">#</span>
|
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
|
40 27 <span class="c">#</span>
|
|
|
41 28
|
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$FindBin::Bin/../lib"</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">File::Copy</span><span class="sc">;</span>
|
|
|
47 34 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
|
|
|
48 35 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
|
49 36 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
|
50 37 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
|
51 38 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsFileUtil</span><span class="sc">;</span>
|
|
|
52 39 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsBitVector</span><span class="sc">;</span>
|
|
|
53 40 <span class="k">use</span> <span class="w">Fingerprints::FingerprintsVector</span><span class="sc">;</span>
|
|
|
54 41
|
|
|
55 42 <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>
|
|
|
56 43
|
|
|
57 44 <span class="c"># Autoflush STDOUT</span>
|
|
|
58 45 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
|
59 46
|
|
|
60 47 <span class="c"># Starting message...</span>
|
|
|
61 48 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
|
62 49 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span>
|
|
|
63 50 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
|
64 51
|
|
|
65 52 <span class="c"># Get the options and setup script...</span>
|
|
|
66 53 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
67 54 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
|
68 55 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">"$FindBin::Bin/$ScriptName"</span><span class="s">)</span><span class="sc">;</span>
|
|
|
69 56 <span class="s">}</span>
|
|
|
70 57
|
|
|
71 58 <span class="k">my</span><span class="s">(</span><span class="i">@FingerprintsFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
|
72 59 <span class="i">@FingerprintsFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">"sdf sd fpf fp csv tsv"</span><span class="s">)</span><span class="sc">;</span>
|
|
|
73 60
|
|
|
74 61 <span class="c"># Process options...</span>
|
|
|
75 62 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
|
76 63 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
|
77 64 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
78 65
|
|
|
79 66 <span class="c"># Setup information about input files...</span>
|
|
|
80 67 <span class="k">print</span> <span class="q">"Checking input fingerprints file(s)...\n"</span><span class="sc">;</span>
|
|
|
81 68 <span class="k">my</span><span class="s">(</span><span class="i">%FingerprintsFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
|
82 69 <span class="i">RetrieveFingerprintsFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
83 70
|
|
|
84 71 <span class="c"># Process input files..</span>
|
|
|
85 72 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
86 73 <span class="k">if</span> <span class="s">(</span><span class="i">@FingerprintsFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
|
87 74 <span class="k">print</span> <span class="q">"\nProcessing fingerprints files...\n"</span><span class="sc">;</span>
|
|
|
88 75 <span class="s">}</span>
|
|
|
89 76 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#FingerprintsFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
|
90 77 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
|
91 78 <span class="k">print</span> <span class="q">"\nProcessing file $FingerprintsFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
|
92 79 <span class="i">GenerateSimilarityMatrices</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
93 80 <span class="s">}</span>
|
|
|
94 81 <span class="s">}</span>
|
|
|
95 82 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
|
96 83
|
|
|
97 84 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
|
98 85 <span class="i">$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>
|
|
|
99 86 <span class="k">print</span> <span class="q">"Total time: "</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
100 87
|
|
|
101 88 <span class="c">###############################################################################</span>
|
|
|
102 89
|
|
|
103 90 <span class="c"># Generate similarity matrices using fingerprints data in text file...</span>
|
|
|
104 91 <span class="c">#</span>
|
|
|
105 <a name="GenerateSimilarityMatrices-"></a> 92 <span class="k">sub </span><span class="m">GenerateSimilarityMatrices</span> <span class="s">{</span>
|
|
|
106 93 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
107 94
|
|
|
108 95 <span class="i">ProcessFingerprintsData</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
109 96
|
|
|
110 97 <span class="k">if</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsBitVectorStringMode</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
|
111 98 <span class="i">GenerateSimilarityMatricesForFingerprintsBitVectors</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
112 99 <span class="s">}</span>
|
|
|
113 100 <span class="k">elsif</span> <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsVectorStringMode</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
|
114 101 <span class="i">GenerateSimilarityMatricesForFingerprintsVectors</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
115 102 <span class="s">}</span>
|
|
|
116 103
|
|
|
117 104 <span class="i">CleanupFingerprintsData</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
118 105 <span class="s">}</span>
|
|
|
119 106
|
|
|
120 107 <span class="c"># Generate bit vector similarity matrices...</span>
|
|
|
121 108 <span class="c">#</span>
|
|
|
122 <a name="GenerateSimilarityMatricesForFingerprintsBitVectors-"></a> 109 <span class="k">sub </span><span class="m">GenerateSimilarityMatricesForFingerprintsBitVectors</span> <span class="s">{</span>
|
|
|
123 110 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
124 111 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$SimilarityMatrixRef</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">@MethodParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
125 112
|
|
|
126 113 <span class="k">for</span> <span class="i">$SpecifiedComparisonMeasure</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
127 114 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsNameRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
128 115 <span class="i">$NewTextFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>] . <span class="q">"${ComparisonMeasure}."</span> . <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
|
129 116
|
|
|
130 117 <span class="i">$MethodName</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsMethodRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
131 118
|
|
|
132 119 <span class="i">@MethodParameters</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
133 120 <span class="i">@MethodParameters</span> = <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsParameterRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>}}<span class="sc">;</span>
|
|
|
134 121
|
|
|
135 122 <span class="i">GenerateSimilarityMatrix</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> \<span class="i">@MethodParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
136 123 <span class="s">}</span>
|
|
|
137 124 <span class="s">}</span>
|
|
|
138 125
|
|
|
139 126 <span class="c"># Generate vector similarity and/or distance matrices...</span>
|
|
|
140 127 <span class="c">#</span>
|
|
|
141 <a name="GenerateSimilarityMatricesForFingerprintsVectors-"></a> 128 <span class="k">sub </span><span class="m">GenerateSimilarityMatricesForFingerprintsVectors</span> <span class="s">{</span>
|
|
|
142 129 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
143 130 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="cm">,</span> <span class="i">$ComparisonMode</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">@MethodParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
144 131
|
|
|
145 132 <span class="k">for</span> <span class="i">$SpecifiedComparisonMeasure</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
146 133 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsNameRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
147 134
|
|
|
148 135 <span class="k">for</span> <span class="i">$ComparisonMode</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonModesRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
149 136 <span class="i">$NewTextFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>] . <span class="q">"${ComparisonMeasure}${ComparisonMode}."</span> . <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
|
150 137
|
|
|
151 138 <span class="i">$MethodName</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsMethodRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
152 139
|
|
|
153 140 <span class="i">@MethodParameters</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
154 141 <span class="k">push</span> <span class="i">@MethodParameters</span><span class="cm">,</span> <span class="i">$ComparisonMode</span><span class="sc">;</span>
|
|
|
155 142 <span class="k">push</span> <span class="i">@MethodParameters</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsParameterRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="s">)</span>}}<span class="sc">;</span>
|
|
|
156 143
|
|
|
157 144 <span class="i">GenerateSimilarityMatrix</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> \<span class="i">@MethodParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
158 145 <span class="s">}</span>
|
|
|
159 146 <span class="s">}</span>
|
|
|
160 147 <span class="s">}</span>
|
|
|
161 148
|
|
|
162 149 <span class="c"># Calculate similarity matrix and write it out...</span>
|
|
|
163 150 <span class="c">#</span>
|
|
|
164 <a name="GenerateSimilarityMatrix-"></a> 151 <span class="k">sub </span><span class="m">GenerateSimilarityMatrix</span> <span class="s">{</span>
|
|
|
165 152 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">$MethodParametersRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
166 153
|
|
|
167 154 <span class="k">print</span> <span class="q">"\nGenerating $NewTextFile...\n"</span><span class="sc">;</span>
|
|
|
168 155
|
|
|
169 156 <span class="c"># Open new file and write out column labels...</span>
|
|
|
170 157 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$NewTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewTextFile: $! \n"</span><span class="sc">;</span>
|
|
|
171 158 <span class="i">WriteColumnLabels</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> \<span class="i">*NEWTEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
|
172 159
|
|
|
173 160 <span class="c"># Calculate and write out similarity matrix values...</span>
|
|
|
174 161 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^LoadInMemory$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
175 162 <span class="i">GenerateSimilarityMatrixUsingMemoryData</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> \<span class="i">*NEWTEXTFILE</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">$MethodParametersRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
176 163 <span class="s">}</span>
|
|
|
177 164 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^ScanFile$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
178 165 <span class="i">GenerateSimilarityMatrixUsingFileData</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> \<span class="i">*NEWTEXTFILE</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">$MethodParametersRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
179 166 <span class="s">}</span>
|
|
|
180 167 <span class="k">else</span> <span class="s">{</span>
|
|
|
181 168 <span class="k">warn</span> <span class="q">"Warning: Input data mode, $OptionsInfo{InputDataMode}, is not supported.\n"</span><span class="sc">;</span>
|
|
|
182 169 <span class="s">}</span>
|
|
|
183 170
|
|
|
184 171 <span class="c"># Close new text file...</span>
|
|
|
185 172 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
|
186 173
|
|
|
187 174 <span class="s">}</span>
|
|
|
188 175
|
|
|
189 176 <span class="c"># Calculate and write out similarity values using fingerprints data already loaded in</span>
|
|
|
190 177 <span class="c"># memory...</span>
|
|
|
191 178 <span class="c">#</span>
|
|
|
192 <a name="GenerateSimilarityMatrixUsingMemoryData-"></a> 179 <span class="k">sub </span><span class="m">GenerateSimilarityMatrixUsingMemoryData</span> <span class="s">{</span>
|
|
|
193 180 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">$MethodParametersRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
194 181 <span class="k">my</span><span class="s">(</span><span class="i">$RowIndex</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="cm">,</span> <span class="i">$CmpdID1</span><span class="cm">,</span> <span class="i">$CmpdID2</span><span class="cm">,</span> <span class="i">$FingerprintsObject1</span><span class="cm">,</span> <span class="i">$FingerprintsObject2</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
|
195 182
|
|
|
196 183 <span class="k">for</span> <span class="i">$RowIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
197 184 <span class="i">$FingerprintsObject1</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>}->[<span class="i">$RowIndex</span>]<span class="sc">;</span>
|
|
|
198 185 <span class="i">$CmpdID1</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>}->[<span class="i">$RowIndex</span>]<span class="sc">;</span>
|
|
|
199 186
|
|
|
200 187 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
201 188 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutQuoteValue}${CmpdID1}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
202 189 <span class="s">}</span>
|
|
|
203 190
|
|
|
204 191 <span class="j">COLINDEX:</span> <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
205 192 <span class="k">if</span> <span class="s">(</span><span class="i">SkipMatrixData</span><span class="s">(</span><span class="i">$RowIndex</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
206 193 <span class="k">next</span> <span class="j">COLINDEX</span><span class="sc">;</span>
|
|
|
207 194 <span class="s">}</span>
|
|
|
208 195
|
|
|
209 196 <span class="i">$FingerprintsObject2</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>}->[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
|
210 197
|
|
|
211 198 <span class="i">$Value</span> = <span class="i">$FingerprintsObject1</span><span class="i">->$MethodName</span><span class="s">(</span><span class="i">$FingerprintsObject2</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$MethodParametersRef</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
212 199 <span class="i">$Value</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">''</span><span class="sc">;</span>
|
|
|
213 200
|
|
|
214 201 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
215 202 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutDelim}$OptionsInfo{OutQuoteValue}${Value}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
216 203 <span class="s">}</span>
|
|
|
217 204 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteIDPairsAndValue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
218 205 <span class="i">$CmpdID2</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>}->[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
|
219 206
|
|
|
220 207 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
221 208 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="s">(</span><span class="i">$CmpdID1</span><span class="cm">,</span> <span class="i">$CmpdID2</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span><span class="sc">;</span>
|
|
|
222 209 <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">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
223 210 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
|
224 211 <span class="s">}</span>
|
|
|
225 212 <span class="s">}</span>
|
|
|
226 213 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
227 214 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
228 215 <span class="s">}</span>
|
|
|
229 216 <span class="s">}</span>
|
|
|
230 217 <span class="s">}</span>
|
|
|
231 218
|
|
|
232 219 <span class="c"># Calculate and write out similarity values by retrieving and prcessing data</span>
|
|
|
233 220 <span class="c"># from fingerprint file...</span>
|
|
|
234 221 <span class="c">#</span>
|
|
|
235 <a name="GenerateSimilarityMatrixUsingFileData-"></a> 222 <span class="k">sub </span><span class="m">GenerateSimilarityMatrixUsingFileData</span> <span class="s">{</span>
|
|
|
236 223 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="cm">,</span> <span class="i">$MethodParametersRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
237 224 <span class="k">my</span><span class="s">(</span><span class="i">$RowIndex</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="cm">,</span> <span class="i">$FingerprintsFileIO</span><span class="cm">,</span> <span class="i">$TmpFingerprintsFileIO</span><span class="cm">,</span> <span class="i">$FingerprintsObject1</span><span class="cm">,</span> <span class="i">$FingerprintsObject2</span><span class="cm">,</span> <span class="i">$CmpdID1</span><span class="cm">,</span> <span class="i">$CmpdID2</span><span class="cm">,</span> <span class="i">$FingerprintsCount</span><span class="cm">,</span> <span class="i">$IgnoredFingerprintsCount</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
|
238 225
|
|
|
239 226 <span class="k">print</span> <span class="q">"\nReading and processing fingerprints data...\n"</span><span class="sc">;</span>
|
|
|
240 227
|
|
|
241 228 <span class="i">$FingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsFileIOParameters</span>}[<span class="i">$FileIndex</span>]}<span class="s">)</span><span class="sc">;</span>
|
|
|
242 229 <span class="i">$FingerprintsFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
243 230
|
|
|
244 231 <span class="i">$RowIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$ColIndex</span> = <span class="n">0</span><span class="sc">;</span>
|
|
|
245 232 <span class="i">$FingerprintsCount</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$IgnoredFingerprintsCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
|
246 233
|
|
|
247 234 <span class="j">FINGERPRINTSFILEIO:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="i">->Read</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
248 235 <span class="i">$FingerprintsCount</span>++<span class="sc">;</span>
|
|
|
249 236
|
|
|
250 237 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="i">->IsFingerprintsDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
251 238 <span class="i">$IgnoredFingerprintsCount</span>++<span class="sc">;</span>
|
|
|
252 239 <span class="k">next</span> <span class="j">FINGERPRINTSFILEIO</span><span class="sc">;</span>
|
|
|
253 240 <span class="s">}</span>
|
|
|
254 241 <span class="i">$RowIndex</span>++<span class="sc">;</span>
|
|
|
255 242 <span class="i">$FingerprintsObject1</span> = <span class="i">$FingerprintsFileIO</span><span class="i">->GetFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
256 243 <span class="i">$CmpdID1</span> = <span class="i">$FingerprintsFileIO</span><span class="i">->GetCompoundID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
257 244
|
|
|
258 245 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
259 246 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutQuoteValue}${CmpdID1}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
260 247 <span class="s">}</span>
|
|
|
261 248
|
|
|
262 249 <span class="c"># Force detail level of 1 to avoid duplicate printing of diagnostic messages for invalid</span>
|
|
|
263 250 <span class="c"># fingerprints data...</span>
|
|
|
264 251 <span class="i">$TmpFingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFileIOParameters</span>}[<span class="i">$FileIndex</span>]}<span class="cm">,</span> <span class="q">"DetailLevel"</span> <span class="cm">=></span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
|
265 252 <span class="i">$TmpFingerprintsFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
266 253
|
|
|
267 254 <span class="i">$ColIndex</span> = <span class="n">0</span><span class="sc">;</span>
|
|
|
268 255 <span class="j">TMPFINGERPRINTSFILEIO:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$TmpFingerprintsFileIO</span><span class="i">->Read</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
269 256 <span class="k">if</span> <span class="s">(</span>!<span class="i">$TmpFingerprintsFileIO</span><span class="i">->IsFingerprintsDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
270 257 <span class="k">next</span> <span class="j">TMPFINGERPRINTSFILEIO</span><span class="sc">;</span>
|
|
|
271 258 <span class="s">}</span>
|
|
|
272 259 <span class="i">$ColIndex</span>++<span class="sc">;</span>
|
|
|
273 260
|
|
|
274 261 <span class="k">if</span> <span class="s">(</span><span class="i">SkipMatrixData</span><span class="s">(</span><span class="i">$RowIndex</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
275 262 <span class="k">next</span> <span class="j">TMPFINGERPRINTSFILEIO</span><span class="sc">;</span>
|
|
|
276 263 <span class="s">}</span>
|
|
|
277 264
|
|
|
278 265 <span class="i">$FingerprintsObject2</span> = <span class="i">$TmpFingerprintsFileIO</span><span class="i">->GetFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
279 266
|
|
|
280 267 <span class="i">$Value</span> = <span class="i">$FingerprintsObject1</span><span class="i">->$MethodName</span><span class="s">(</span><span class="i">$FingerprintsObject2</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$MethodParametersRef</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
281 268 <span class="i">$Value</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">''</span><span class="sc">;</span>
|
|
|
282 269
|
|
|
283 270 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
284 271 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutDelim}$OptionsInfo{OutQuoteValue}${Value}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
285 272 <span class="s">}</span>
|
|
|
286 273 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteIDPairsAndValue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
287 274 <span class="i">$CmpdID2</span> = <span class="i">$TmpFingerprintsFileIO</span><span class="i">->GetCompoundID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
288 275
|
|
|
289 276 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
290 277 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="s">(</span><span class="i">$CmpdID1</span><span class="cm">,</span> <span class="i">$CmpdID2</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span><span class="sc">;</span>
|
|
|
291 278 <span class="i">$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>
|
|
|
292 279 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
|
293 280 <span class="s">}</span>
|
|
|
294 281 <span class="s">}</span>
|
|
|
295 282 <span class="i">$TmpFingerprintsFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
296 283
|
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
298 285 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
299 286 <span class="s">}</span>
|
|
|
300 287 <span class="s">}</span>
|
|
|
301 288
|
|
|
302 289 <span class="i">$FingerprintsFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
303 290
|
|
|
304 291 <span class="k">print</span> <span class="q">"Number of fingerprints data entries in database fingerprints file: $FingerprintsCount\n"</span><span class="sc">;</span>
|
|
|
305 292 <span class="k">print</span> <span class="q">"Number of fingerprints date entries processed successfully: "</span><span class="cm">,</span> <span class="s">(</span><span class="i">$FingerprintsCount</span> - <span class="i">$IgnoredFingerprintsCount</span><span class="s">)</span> <span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
306 293 <span class="k">print</span> <span class="q">"Number of fingerprints data entries ignored due to missing/invalid data: $IgnoredFingerprintsCount\n\n"</span><span class="sc">;</span>
|
|
|
307 294 <span class="s">}</span>
|
|
|
308 295
|
|
|
309 296 <span class="c"># Check whether matrix data need to be skipped...</span>
|
|
|
310 297 <span class="c">#</span>
|
|
|
311 <a name="SkipMatrixData-"></a> 298 <span class="k">sub </span><span class="m">SkipMatrixData</span> <span class="s">{</span>
|
|
|
312 299 <span class="k">my</span><span class="s">(</span><span class="i">$RowIndex</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
313 300
|
|
|
314 301 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteFullMatrix</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
315 302 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
316 303 <span class="s">}</span>
|
|
|
317 304 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteUpperTriangularMatrix</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
318 305 <span class="k">return</span> <span class="s">(</span><span class="i">$RowIndex</span> > <span class="i">$ColIndex</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
319 306 <span class="s">}</span>
|
|
|
320 307 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">WriteLowerTriangularMatrix</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
321 308 <span class="k">return</span> <span class="s">(</span><span class="i">$RowIndex</span> < <span class="i">$ColIndex</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
322 309 <span class="s">}</span>
|
|
|
323 310
|
|
|
324 311 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
325 312 <span class="s">}</span>
|
|
|
326 313
|
|
|
327 314 <span class="c"># Write out column labels...</span>
|
|
|
328 315 <span class="c">#</span>
|
|
|
329 <a name="WriteColumnLabels-"></a> 316 <span class="k">sub </span><span class="m">WriteColumnLabels</span> <span class="s">{</span>
|
|
|
330 317 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
331 318 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
|
332 319
|
|
|
333 320 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixFormat</span>} =~ <span class="q">/^IDPairsAndValue$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
334 321 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
335 322 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="s">(</span><span class="q">'CmpdID1'</span><span class="cm">,</span> <span class="q">'CmpdID2'</span><span class="cm">,</span> <span class="q">'Coefficient Value'</span><span class="s">)</span><span class="sc">;</span>
|
|
|
336 323 <span class="i">$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>
|
|
|
337 324 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
|
338 325 <span class="s">}</span>
|
|
|
339 326 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixFormat</span>} =~ <span class="q">/^RowsAndColumns$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
340 327 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^LoadInMemory$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
341 328 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
342 329 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="q">''</span><span class="sc">;</span>
|
|
|
343 330 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>}}<span class="sc">;</span>
|
|
|
344 331 <span class="i">$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>
|
|
|
345 332 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
|
346 333 <span class="s">}</span>
|
|
|
347 334 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^ScanFile$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
348 335 <span class="k">my</span><span class="s">(</span> <span class="i">$FingerprintsFileIO</span><span class="cm">,</span> <span class="i">$CmpdID</span><span class="s">)</span><span class="sc">;</span>
|
|
|
349 336
|
|
|
350 337 <span class="c"># Scan file to retrieve compound IDs...</span>
|
|
|
351 338 <span class="c">#</span>
|
|
|
352 339 <span class="k">print</span> <span class="q">"\nProcessing fingerprints file to generate compound IDs...\n"</span><span class="sc">;</span>
|
|
|
353 340
|
|
|
354 341 <span class="c"># Force detail level of 1 to avoid diagnostics messages for invalid fingeprints data during</span>
|
|
|
355 342 <span class="c"># retrieval of compound IDs as these get printed out during calculation of matrix...</span>
|
|
|
356 343 <span class="c">#</span>
|
|
|
357 344 <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">FingerprintsFileIOParameters</span>}[<span class="i">$FileIndex</span>]}<span class="cm">,</span> <span class="q">"DetailLevel"</span> <span class="cm">=></span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
|
358 345 <span class="i">$FingerprintsFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
359 346
|
|
|
360 347 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutQuoteValue}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
361 348
|
|
|
362 349 <span class="j">FINGERPRINTSFILEIO:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="i">->Read</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
363 350 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="i">->IsFingerprintsDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
364 351 <span class="k">next</span> <span class="j">FINGERPRINTSFILEIO</span><span class="sc">;</span>
|
|
|
365 352 <span class="s">}</span>
|
|
|
366 353 <span class="i">$CmpdID</span> = <span class="i">$FingerprintsFileIO</span><span class="i">->GetCompoundID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
367 354 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$OptionsInfo{OutDelim}$OptionsInfo{OutQuoteValue}${CmpdID}$OptionsInfo{OutQuoteValue}"</span><span class="sc">;</span>
|
|
|
368 355 <span class="s">}</span>
|
|
|
369 356 <span class="i">$FingerprintsFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
370 357
|
|
|
371 358 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
372 359
|
|
|
373 360 <span class="k">print</span> <span class="q">"Processing fingerprints file to generate matrix...\n"</span><span class="sc">;</span>
|
|
|
374 361 <span class="s">}</span>
|
|
|
375 362 <span class="s">}</span>
|
|
|
376 363 <span class="k">else</span> <span class="s">{</span>
|
|
|
377 364 <span class="k">warn</span> <span class="q">"Warning: Output matrix format, $OptionsInfo{OutMatrixFormat}, is not supported.\n"</span><span class="sc">;</span>
|
|
|
378 365 <span class="s">}</span>
|
|
|
379 366 <span class="s">}</span>
|
|
|
380 367
|
|
|
381 368 <span class="c"># Process fingerprints data...</span>
|
|
|
382 369 <span class="c">#</span>
|
|
|
383 <a name="ProcessFingerprintsData-"></a> 370 <span class="k">sub </span><span class="m">ProcessFingerprintsData</span> <span class="s">{</span>
|
|
|
384 371 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
385 372 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
|
|
|
386 373
|
|
|
387 374 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
|
388 375 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
|
389 376
|
|
|
390 377 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^LoadInMemory$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
391 378 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
|
|
|
392 379
|
|
|
393 380 <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">FingerprintsFileIOParameters</span>}[<span class="i">$FileIndex</span>]}<span class="s">)</span><span class="sc">;</span>
|
|
|
394 381 <span class="s">(</span><span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>}<span class="cm">,</span> <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>}<span class="s">)</span> = <span class="i">Fingerprints::FingerprintsFileUtil::ReadAndProcessFingerpritsData</span><span class="s">(</span><span class="i">$FingerprintsFileIO</span><span class="s">)</span><span class="sc">;</span>
|
|
|
395 382 <span class="s">}</span>
|
|
|
396 383 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^ScanFile$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
397 384 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$TmpFingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
|
|
|
398 385
|
|
|
399 386 <span class="i">$FingerprintsFile</span> = <span class="i">$FingerprintsFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
|
400 387 <span class="i">$TmpFingerprintsFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFile</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
|
401 388
|
|
|
402 389 <span class="c"># Copy fingerprints file to a tmp file for calculating similarity matrix...</span>
|
|
|
403 390 <span class="k">print</span> <span class="q">"\nCopying fingerprints file, $FingerprintsFile, to temporary fingperints file, $TmpFingerprintsFile...\n"</span><span class="sc">;</span>
|
|
|
404 391 <span class="w">copy</span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$TmpFingerprintsFile</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't copy $FingerprintsFile to $TmpFingerprintsFile: $! \n"</span><span class="sc">;</span>
|
|
|
405 392 <span class="s">}</span>
|
|
|
406 393 <span class="s">}</span>
|
|
|
407 394
|
|
|
408 395 <span class="c"># Clean up fingerprints data...</span>
|
|
|
409 396 <span class="c">#</span>
|
|
|
410 <a name="CleanupFingerprintsData-"></a> 397 <span class="k">sub </span><span class="m">CleanupFingerprintsData</span> <span class="s">{</span>
|
|
|
411 398 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
412 399
|
|
|
413 400 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^LoadInMemory$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
414 401 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">CompundIDsRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
|
415 402 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsObjectsRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
|
416 403 <span class="s">}</span>
|
|
|
417 404 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} =~ <span class="q">/^ScanFile$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
418 405 <span class="k">my</span><span class="s">(</span><span class="i">$TmpFingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
|
|
|
419 406
|
|
|
420 407 <span class="c"># Delete temporary fingerprints file...</span>
|
|
|
421 408 <span class="i">$TmpFingerprintsFile</span> = <span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFile</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
|
422 409
|
|
|
423 410 <span class="k">print</span> <span class="q">"\nDeleting temporary fingerprints file $TmpFingerprintsFile...\n"</span><span class="sc">;</span>
|
|
|
424 411 <span class="k">unlink</span> <span class="i">$TmpFingerprintsFile</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't unlink $TmpFingerprintsFile: $! \n"</span><span class="sc">;</span>
|
|
|
425 412 <span class="s">}</span>
|
|
|
426 413 <span class="s">}</span>
|
|
|
427 414
|
|
|
428 415 <span class="c"># Retrieve information about fingerprints files...</span>
|
|
|
429 416 <span class="c">#</span>
|
|
|
430 <a name="RetrieveFingerprintsFilesInfo-"></a> 417 <span class="k">sub </span><span class="m">RetrieveFingerprintsFilesInfo</span> <span class="s">{</span>
|
|
|
431 418 <span class="k">my</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="i">$TmpFingerprintsFile</span><span class="cm">,</span> <span class="i">$FingerprintsFileIO</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">$FileType</span><span class="cm">,</span> <span class="i">$Index</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">$FileName</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$OutFileExt</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
432 419
|
|
|
433 420 <span class="i">%FingerprintsFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
434 421 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
435 422 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileType</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
436 423 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">InDelim</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
437 424 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
438 425 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileExt</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
439 426
|
|
|
440 427 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFile</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
441 428
|
|
|
442 429 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsFileIOParameters</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
443 430 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFileIOParameters</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
444 431
|
|
|
445 432 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsBitVectorStringMode</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
446 433 <span class="i">@</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsVectorStringMode</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
447 434
|
|
|
448 435 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#FingerprintsFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
|
449 436 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
|
450 437 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileType</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
|
451 438 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
|
452 439 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
|
453 440 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
|
454 441
|
|
|
455 442 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsFileIOParameters</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
456 443
|
|
|
457 444 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFile</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
|
458 445 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFileIOParameters</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
459 446
|
|
|
460 447 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsBitVectorStringMode</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
|
461 448 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsVectorStringMode</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
|
462 449
|
|
|
463 450 <span class="i">$FingerprintsFile</span> = <span class="i">$FingerprintsFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
|
464 451 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$FingerprintsFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
465 452 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $FingerprintsFile: It doesn't exist\n"</span><span class="sc">;</span>
|
|
|
466 453 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
467 454 <span class="s">}</span>
|
|
|
468 455
|
|
|
469 456 <span class="i">$FileType</span> = <span class="i">Fingerprints::FingerprintsFileUtil::GetFingerprintsFileType</span><span class="s">(</span><span class="i">$FingerprintsFile</span><span class="s">)</span><span class="sc">;</span>
|
|
|
470 457 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$FileType</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
471 458 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $FingerprintsFile: It's not a fingerprints file\n"</span><span class="sc">;</span>
|
|
|
472 459 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
473 460 <span class="s">}</span>
|
|
|
474 461
|
|
|
475 462 <span class="i">$FileDir</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
|
476 463 <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>
|
|
|
477 464
|
|
|
478 465 <span class="c"># Setup temporary fingerprints file name for scan file mode...</span>
|
|
|
479 466 <span class="i">$TmpFingerprintsFile</span> = <span class="q">"${FileName}Tmp.${FileExt}"</span><span class="sc">;</span>
|
|
|
480 467
|
|
|
481 468 <span class="i">$InDelim</span> = <span class="s">(</span><span class="i">$FileExt</span> =~ <span class="q">/^tsv$/i</span><span class="s">)</span> ? <span class="q">'Tab'</span> <span class="co">:</span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="sc">;</span>
|
|
|
482 469
|
|
|
483 470 <span class="c"># Setup output file names...</span>
|
|
|
484 471 <span class="i">$OutFileExt</span> = <span class="q">"csv"</span><span class="sc">;</span>
|
|
|
485 472 <span class="k">if</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="s">{</span>
|
|
|
486 473 <span class="i">$OutFileExt</span> = <span class="q">"tsv"</span><span class="sc">;</span>
|
|
|
487 474 <span class="s">}</span>
|
|
|
488 475
|
|
|
489 476 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
|
490 477 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} && <span class="s">(</span><span class="i">@FingerprintsFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
491 478 <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>
|
|
|
492 479 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
|
493 480 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
|
494 481 <span class="s">}</span>
|
|
|
495 482 <span class="k">else</span> <span class="s">{</span>
|
|
|
496 483 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
|
|
|
497 484 <span class="s">}</span>
|
|
|
498 485 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
|
499 486 <span class="s">}</span>
|
|
|
500 487
|
|
|
501 488 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">overwrite</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
502 489 <span class="c"># Similarity matrices output file names for bit-vector strings...</span>
|
|
|
503 490 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="s">)</span><span class="sc">;</span>
|
|
|
504 491 <span class="k">for</span> <span class="i">$SpecifiedComparisonMeasure</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
505 492 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsNameRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
506 493 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}${ComparisonMeasure}.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
|
507 494 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $FingerprintsFile: The file ${OutFileRoot}${ComparisonMeasure}.${OutFileExt} already exists.\n"</span><span class="sc">;</span>
|
|
|
508 495 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
509 496 <span class="s">}</span>
|
|
|
510 497 <span class="s">}</span>
|
|
|
511 498 <span class="c"># Similarity matrices output file names for vector strings...</span>
|
|
|
512 499 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMode</span><span class="s">)</span><span class="sc">;</span>
|
|
|
513 500 <span class="k">for</span> <span class="i">$SpecifiedComparisonMeasure</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
514 501 <span class="i">$ComparisonMeasure</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsNameRef</span>}->{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedComparisonMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
515 502 <span class="k">for</span> <span class="i">$ComparisonMode</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonModesRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
516 503 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}${ComparisonMeasure}${ComparisonMode}.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
|
517 504 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $FingerprintsFile: The file ${OutFileRoot}${ComparisonMeasure}${ComparisonMode}.${OutFileExt} already exists.\n"</span><span class="sc">;</span>
|
|
|
518 505 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
519 506 <span class="s">}</span>
|
|
|
520 507 <span class="s">}</span>
|
|
|
521 508 <span class="s">}</span>
|
|
|
522 509 <span class="s">}</span>
|
|
|
523 510
|
|
|
524 511 <span class="c"># Setup FingerprintsFileIO parameters...</span>
|
|
|
525 512 <span class="i">%FingerprintsFileIOParameters</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
526 513 <span class="j">FILEIOPARAMETERS:</span> <span class="s">{</span>
|
|
|
527 514 <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>
|
|
|
528 515 <span class="i">%FingerprintsFileIOParameters</span> = <span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">'Mode'</span> <span class="cm">=></span> <span class="q">'Read'</span><span class="cm">,</span> <span class="q">'FingerprintsStringMode'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>}<span class="cm">,</span> <span class="q">'ValidateData'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">'DetailLevel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="cm">,</span> <span class="q">'FingerprintsFieldLabel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsField</span>}<span class="cm">,</span> <span class="q">'CompoundIDMode'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>}<span class="cm">,</span> <span class="q">'CompoundIDFieldLabel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDField</span>}<span class="cm">,</span> <span class="q">'CompoundIDPrefix'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDPrefix</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
529 516 <span class="k">last</span> <span class="j">FILEIOPARAMETERS</span><span class="sc">;</span>
|
|
|
530 517 <span class="s">}</span>
|
|
|
531 518 <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>
|
|
|
532 519 <span class="i">%FingerprintsFileIOParameters</span> = <span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">'Mode'</span> <span class="cm">=></span> <span class="q">'Read'</span><span class="cm">,</span> <span class="q">'FingerprintsStringMode'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>}<span class="cm">,</span> <span class="q">'ValidateData'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">'DetailLevel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
533 520 <span class="k">last</span> <span class="j">FILEIOPARAMETERS</span><span class="sc">;</span>
|
|
|
534 521 <span class="s">}</span>
|
|
|
535 522 <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>
|
|
|
536 523 <span class="i">%FingerprintsFileIOParameters</span> = <span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$FingerprintsFile</span><span class="cm">,</span> <span class="q">'Mode'</span> <span class="cm">=></span> <span class="q">'Read'</span><span class="cm">,</span> <span class="q">'FingerprintsStringMode'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>}<span class="cm">,</span> <span class="q">'ValidateData'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">ValidateData</span>}<span class="cm">,</span> <span class="q">'DetailLevel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>}<span class="cm">,</span> <span class="q">'FingerprintsCol'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsCol</span>}<span class="cm">,</span> <span class="q">'ColMode'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">ColMode</span>}<span class="cm">,</span> <span class="q">'CompoundIDCol'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDCol</span>}<span class="cm">,</span> <span class="q">'CompoundIDPrefix'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDPrefix</span>}<span class="cm">,</span> <span class="q">'InDelim'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
537 524 <span class="k">last</span> <span class="j">FILEIOPARAMETERS</span><span class="sc">;</span>
|
|
|
538 525 <span class="s">}</span>
|
|
|
539 526 <span class="k">warn</span> <span class="q">"Warning: File type for fingerprints file, $FingerprintsFile, is not valid. Supported file types: SD, FP or Text\n"</span><span class="sc">;</span>
|
|
|
540 527 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
541 528 <span class="s">}</span>
|
|
|
542 529
|
|
|
543 530 <span class="c"># Retrieve fingerints file string mode information...</span>
|
|
|
544 531 <span class="i">$FingerprintsFileIO</span> = <span class="i">Fingerprints::FingerprintsFileUtil::NewFingerprintsFileIO</span><span class="s">(</span><span class="i">%FingerprintsFileIOParameters</span><span class="s">)</span><span class="sc">;</span>
|
|
|
545 532
|
|
|
546 533 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
|
547 534 <span class="k">warn</span> <span class="q">"Warning: Ignoring fingerprints file $FingerprintsFile: It contains invalid fingerprints data\n"</span><span class="sc">;</span>
|
|
|
548 535 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
549 536 <span class="s">}</span>
|
|
|
550 537 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FingerprintsFileIO</span><span class="i">->IsFingerprintsFileDataValid</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
551 538 <span class="k">warn</span> <span class="q">"Warning: Ignoring fingerprints file $FingerprintsFile: It contains invalid fingerprints data\n"</span><span class="sc">;</span>
|
|
|
552 539 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
|
553 540 <span class="s">}</span>
|
|
|
554 541 <span class="i">$FingerprintsBitVectorStringMode</span> = <span class="i">$FingerprintsFileIO</span><span class="i">->GetFingerprintsBitVectorStringMode</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
555 542 <span class="i">$FingerprintsVectorStringMode</span> = <span class="i">$FingerprintsFileIO</span><span class="i">->GetFingerprintsVectorStringMode</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
556 543
|
|
|
557 544
|
|
|
558 545 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
|
559 546 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FileType</span>}[<span class="i">$Index</span>] = <span class="i">$FileType</span><span class="sc">;</span>
|
|
|
560 547
|
|
|
561 548 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="i">$InDelim</span><span class="sc">;</span>
|
|
|
562 549
|
|
|
563 550 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span>
|
|
|
564 551 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileExt</span><span class="sc">;</span>
|
|
|
565 552
|
|
|
566 553 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsFileIOParameters</span>}[<span class="i">$Index</span>]} = <span class="i">%FingerprintsFileIOParameters</span><span class="sc">;</span>
|
|
|
567 554
|
|
|
568 555 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFile</span>}[<span class="i">$Index</span>] = <span class="i">$TmpFingerprintsFile</span><span class="sc">;</span>
|
|
|
569 556
|
|
|
570 557 <span class="i">$FingerprintsFileIOParameters</span>{<span class="w">Name</span>} = <span class="i">$TmpFingerprintsFile</span><span class="sc">;</span>
|
|
|
571 558 <span class="i">%</span>{<span class="i">$FingerprintsFilesInfo</span>{<span class="w">TmpFingerprintsFileIOParameters</span>}[<span class="i">$Index</span>]} = <span class="i">%FingerprintsFileIOParameters</span><span class="sc">;</span>
|
|
|
572 559
|
|
|
573 560 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsBitVectorStringMode</span>}[<span class="i">$Index</span>] = <span class="i">$FingerprintsBitVectorStringMode</span><span class="sc">;</span>
|
|
|
574 561 <span class="i">$FingerprintsFilesInfo</span>{<span class="w">FingerprintsVectorStringMode</span>}[<span class="i">$Index</span>] = <span class="i">$FingerprintsVectorStringMode</span><span class="sc">;</span>
|
|
|
575 562 <span class="s">}</span>
|
|
|
576 563 <span class="s">}</span>
|
|
|
577 564
|
|
|
578 565 <span class="c"># Process option values...</span>
|
|
|
579 <a name="ProcessOptions-"></a> 566 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
|
580 567 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
581 568
|
|
|
582 569 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
|
583 570
|
|
|
584 571 <span class="i">$OptionsInfo</span>{<span class="w">InputDataMode</span>} = <span class="i">$Options</span>{<span class="w">inputdatamode</span>}<span class="sc">;</span>
|
|
|
585 572
|
|
|
586 573 <span class="i">ProcessBitVectorComparisonOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
587 574 <span class="i">ProcessVectorComparisonOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
588 575
|
|
|
589 576 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDPrefix</span>} = <span class="i">$Options</span>{<span class="w">compoundidprefix</span>} ? <span class="i">$Options</span>{<span class="w">compoundidprefix</span>} <span class="co">:</span> <span class="q">'Cmpd'</span><span class="sc">;</span>
|
|
|
590 577
|
|
|
591 578 <span class="c"># Compound ID and fingerprints column options for text files...</span>
|
|
|
592 579 <span class="i">$OptionsInfo</span>{<span class="w">ColMode</span>} = <span class="i">$Options</span>{<span class="w">colmode</span>}<span class="sc">;</span>
|
|
|
593 580
|
|
|
594 581 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
595 582 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
596 583 <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">compoundidcol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
597 584 <span class="k">die</span> <span class="q">"Error: Column value, $Options{compoundidcol}, specified using \"--CompoundIDCol\" is not valid: Allowed integer values: > 0\n"</span><span class="sc">;</span>
|
|
|
598 585 <span class="s">}</span>
|
|
|
599 586 <span class="s">}</span>
|
|
|
600 587 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDCol</span>} = <span class="i">$Options</span>{<span class="w">compoundidcol</span>}<span class="sc">;</span>
|
|
|
601 588 <span class="s">}</span>
|
|
|
602 589 <span class="k">else</span> <span class="s">{</span>
|
|
|
603 590 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDCol</span>} = <span class="q">'AutoDetect'</span><span class="sc">;</span>
|
|
|
604 591 <span class="s">}</span>
|
|
|
605 592
|
|
|
606 593 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
607 594 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} =~ <span class="q">/^ColNum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
608 595 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
609 596 <span class="k">die</span> <span class="q">"Error: Column value, $Options{fingerprintscol}, specified using \"--FingerprintsCol\" is not valid: Allowed integer values: > 0\n"</span><span class="sc">;</span>
|
|
|
610 597 <span class="s">}</span>
|
|
|
611 598 <span class="s">}</span>
|
|
|
612 599 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsCol</span>} = <span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="sc">;</span>
|
|
|
613 600 <span class="s">}</span>
|
|
|
614 601 <span class="k">else</span> <span class="s">{</span>
|
|
|
615 602 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsCol</span>} = <span class="q">'AutoDetect'</span><span class="sc">;</span>
|
|
|
616 603 <span class="s">}</span>
|
|
|
617 604
|
|
|
618 605 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidcol</span>}<span class="s">)</span> && <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
619 606 <span class="k">if</span> <span class="s">(</span><span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidcol</span>}<span class="s">)</span> && <span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
620 607 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidcol</span>} == <span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
621 608 <span class="k">die</span> <span class="q">"Error: Values specified using \"--CompoundIDCol\" and \"--FingerprintsCol\", $Options{compoundidcol}, must be different.\n"</span><span class="sc">;</span>
|
|
|
622 609 <span class="s">}</span>
|
|
|
623 610 <span class="s">}</span>
|
|
|
624 611 <span class="k">else</span> <span class="s">{</span>
|
|
|
625 612 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidcol</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">fingerprintscol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
626 613 <span class="k">die</span> <span class="q">"Error: Values specified using \"--CompoundIDCol\" and \"--FingerprintsCol\", $Options{compoundidcol}, must be different.\n"</span><span class="sc">;</span>
|
|
|
627 614 <span class="s">}</span>
|
|
|
628 615 <span class="s">}</span>
|
|
|
629 616 <span class="s">}</span>
|
|
|
630 617
|
|
|
631 618 <span class="c"># Compound ID and fingerprints field options for SD files...</span>
|
|
|
632 619 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} = <span class="i">$Options</span>{<span class="w">compoundidmode</span>}<span class="sc">;</span>
|
|
|
633 620 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDField</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
634 621
|
|
|
635 622 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
636 623 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">compoundidfield</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
637 624 <span class="k">die</span> <span class="q">"Error: You must specify a value for \"--CompoundIDField\" option in \"DataField\" \"--CompoundIDMode\". \n"</span><span class="sc">;</span>
|
|
|
638 625 <span class="s">}</span>
|
|
|
639 626 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDField</span>} = <span class="i">$Options</span>{<span class="w">compoundidfield</span>}<span class="sc">;</span>
|
|
|
640 627 <span class="s">}</span>
|
|
|
641 628
|
|
|
642 629
|
|
|
643 630 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
644 631 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsField</span>} = <span class="i">$Options</span>{<span class="w">fingerprintsfield</span>}<span class="sc">;</span>
|
|
|
645 632 <span class="s">}</span>
|
|
|
646 633 <span class="k">else</span> <span class="s">{</span>
|
|
|
647 634 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsField</span>} = <span class="q">'AutoDetect'</span><span class="sc">;</span>
|
|
|
648 635 <span class="s">}</span>
|
|
|
649 636
|
|
|
650 637 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidfield</span>} && <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
651 638 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidfield</span>} <span class="k">eq</span> <span class="i">$Options</span>{<span class="w">fingerprintsfield</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
652 639 <span class="k">die</span> <span class="q">"Error: Values specified using \"--CompoundIDField\" and \"--Fingerprintsfield\", $Options{compoundidfield}, must be different.\n"</span><span class="sc">;</span>
|
|
|
653 640 <span class="s">}</span>
|
|
|
654 641 <span class="s">}</span>
|
|
|
655 642
|
|
|
656 643 <span class="i">$OptionsInfo</span>{<span class="w">Detail</span>} = <span class="i">$Options</span>{<span class="w">detail</span>}<span class="sc">;</span>
|
|
|
657 644
|
|
|
658 645 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="i">$Options</span>{<span class="w">indelim</span>}<span class="sc">;</span>
|
|
|
659 646 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/tab/i</span> <span class="s">)</span> ? <span class="q">"\t"</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span><span class="s">)</span> ? <span class="q">"\;"</span> <span class="co">:</span> <span class="q">"\,"</span><span class="s">)</span><span class="sc">;</span>
|
|
|
660 647 <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$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
661 648 <span class="i">$OptionsInfo</span>{<span class="w">OutQuoteValue</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="q">'"'</span> <span class="co">:</span> <span class="q">''</span><span class="sc">;</span>
|
|
|
662 649
|
|
|
663 650 <span class="i">$OptionsInfo</span>{<span class="w">OutMatrixFormat</span>} = <span class="i">$Options</span>{<span class="w">outmatrixformat</span>}<span class="sc">;</span>
|
|
|
664 651
|
|
|
665 652 <span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">WriteIDPairsAndValue</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
|
666 653 <span class="j">OUTMATRIXFORMAT:</span> <span class="s">{</span>
|
|
|
667 654 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixFormat</span>} =~ <span class="q">/^RowsAndColumns$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
668 655 <span class="i">$OptionsInfo</span>{<span class="w">WriteRowsAndColumns</span>} = <span class="n">1</span><span class="sc">;</span> <span class="k">last</span> <span class="j">OUTMATRIXFORMAT</span><span class="sc">;</span>
|
|
|
669 656 <span class="s">}</span>
|
|
|
670 657 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixFormat</span>} =~ <span class="q">/^IDPairsAndValue$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
671 658 <span class="i">$OptionsInfo</span>{<span class="w">WriteIDPairsAndValue</span>} = <span class="n">1</span><span class="sc">;</span> <span class="k">last</span> <span class="j">OUTMATRIXFORMAT</span><span class="sc">;</span>
|
|
|
672 659 <span class="s">}</span>
|
|
|
673 660 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outmatrixformat}, for option \"--OutMatrixFormat\" is not valid. Allowed values: RowsAndColumns or IDPairsAndValue\n"</span><span class="sc">;</span>
|
|
|
674 661 <span class="s">}</span>
|
|
|
675 662
|
|
|
676 663 <span class="i">$OptionsInfo</span>{<span class="w">OutMatrixType</span>} = <span class="i">$Options</span>{<span class="w">outmatrixtype</span>}<span class="sc">;</span>
|
|
|
677 664
|
|
|
678 665 <span class="i">$OptionsInfo</span>{<span class="w">WriteFullMatrix</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
|
679 666 <span class="i">$OptionsInfo</span>{<span class="w">WriteUpperTriangularMatrix</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">WriteLowerTriangularMatrix</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
|
680 667 <span class="j">OUTMATRIXTYPE:</span> <span class="s">{</span>
|
|
|
681 668 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixType</span>} =~ <span class="q">/^FullMatrix$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
682 669 <span class="i">$OptionsInfo</span>{<span class="w">WriteFullMatrix</span>} = <span class="n">1</span><span class="sc">;</span> <span class="k">last</span> <span class="j">OUTMATRIXTYPE</span><span class="sc">;</span>
|
|
|
683 670 <span class="s">}</span>
|
|
|
684 671 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixType</span>} =~ <span class="q">/^UpperTriangularMatrix$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
685 672 <span class="i">$OptionsInfo</span>{<span class="w">WriteUpperTriangularMatrix</span>} = <span class="n">1</span><span class="sc">;</span> <span class="k">last</span> <span class="j">OUTMATRIXTYPE</span><span class="sc">;</span>
|
|
|
686 673 <span class="s">}</span>
|
|
|
687 674 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutMatrixType</span>} =~ <span class="q">/^LowerTriangularMatrix$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
688 675 <span class="i">$OptionsInfo</span>{<span class="w">WriteLowerTriangularMatrix</span>} = <span class="n">1</span><span class="sc">;</span> <span class="k">last</span> <span class="j">OUTMATRIXTYPE</span><span class="sc">;</span>
|
|
|
689 676 <span class="s">}</span>
|
|
|
690 677 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outmatrixtype}, for option \"--OutMatrixType\" is not valid. Allowed values: FullMatrix, UpperTriangularMatrix or LowerTriangularMatrix\n"</span><span class="sc">;</span>
|
|
|
691 678 <span class="s">}</span>
|
|
|
692 679
|
|
|
693 680 <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
694 681 <span class="i">$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>
|
|
|
695 682
|
|
|
696 683 <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>
|
|
|
697 684 <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>
|
|
|
698 685
|
|
|
699 686 <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>} = <span class="i">$Options</span>{<span class="w">precision</span>}<span class="sc">;</span>
|
|
|
700 687
|
|
|
701 688 <span class="s">}</span>
|
|
|
702 689
|
|
|
703 690 <span class="c"># Process options related to comparion of bit vector strings...</span>
|
|
|
704 691 <span class="c">#</span>
|
|
|
705 <a name="ProcessBitVectorComparisonOptions-"></a> 692 <span class="k">sub </span><span class="m">ProcessBitVectorComparisonOptions</span> <span class="s">{</span>
|
|
|
706 693 <span class="c"># Setup supported bit vector similarity coefficients for bit vector strings...</span>
|
|
|
707 694 <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>
|
|
|
708 695
|
|
|
709 696 <span class="i">@SupportedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
710 697 <span class="i">%SupportedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
711 698 <span class="i">%SupportedComparisonMeasuresMethodMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
712 699
|
|
|
713 700 <span class="k">for</span> <span class="i">$SupportedComparisonMeasure</span> <span class="s">(</span><span class="i">Fingerprints::FingerprintsBitVector::GetSupportedSimilarityCoefficients</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
714 701 <span class="c"># Similarity coefficient function/method names contain "Coefficient" in their names.</span>
|
|
|
715 702 <span class="c"># So take 'em out and setup a map to original function/method name...</span>
|
|
|
716 703 <span class="i">$ComparisonMeasure</span> = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
|
|
|
717 704 <span class="i">$ComparisonMeasure</span> =~ <span class="q">s/Coefficient$//</span><span class="sc">;</span>
|
|
|
718 705
|
|
|
719 706 <span class="k">push</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
|
|
|
720 707 <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">$ComparisonMeasure</span><span class="sc">;</span>
|
|
|
721 708 <span class="i">$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>
|
|
|
722 709 <span class="s">}</span>
|
|
|
723 710
|
|
|
724 711 <span class="c"># Setup a list of similarity coefficients to use for calculating similarity matrices for bit vector strings...</span>
|
|
|
725 712 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="cm">,</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresNameMap</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresMethodMap</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresParameterMap</span><span class="s">)</span><span class="sc">;</span>
|
|
|
726 713
|
|
|
727 714 <span class="i">@SpecifiedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
728 715 <span class="i">%SpecifiedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
729 716 <span class="i">%SpecifiedComparisonMeasuresMethodMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
730 717 <span class="i">%SpecifiedComparisonMeasuresParameterMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
731 718
|
|
|
732 719 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
733 720 <span class="k">push</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">@SupportedComparisonMeasures</span><span class="sc">;</span>
|
|
|
734 721 <span class="s">}</span>
|
|
|
735 722 <span class="k">else</span> <span class="s">{</span>
|
|
|
736 723 <span class="c"># Comma delimited list of similarity coefficients...</span>
|
|
|
737 724 <span class="k">my</span><span class="s">(</span><span class="i">$BitVectorComparisonMode</span><span class="cm">,</span> <span class="i">@SpecifiedMeasures</span><span class="cm">,</span> <span class="i">@UnsupportedSpecifiedMeasures</span><span class="s">)</span><span class="sc">;</span>
|
|
|
738 725
|
|
|
739 726 <span class="i">$BitVectorComparisonMode</span> = <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>}<span class="sc">;</span>
|
|
|
740 727 <span class="i">$BitVectorComparisonMode</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
|
741 728 <span class="i">@SpecifiedMeasures</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$BitVectorComparisonMode</span><span class="sc">;</span>
|
|
|
742 729 <span class="i">@UnsupportedSpecifiedMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
743 730
|
|
|
744 731 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
745 732 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
746 733 <span class="k">push</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
|
|
|
747 734 <span class="s">}</span>
|
|
|
748 735 <span class="k">else</span> <span class="s">{</span>
|
|
|
749 736 <span class="k">push</span> <span class="i">@UnsupportedSpecifiedMeasures</span><span class="cm">,</span> <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
|
|
|
750 737 <span class="s">}</span>
|
|
|
751 738 <span class="s">}</span>
|
|
|
752 739 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
753 740 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedMeasures</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
|
754 741 <span class="k">warn</span> <span class="q">"Error: The values specified - "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@UnsupportedSpecifiedMeasures</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span><span class="q">" - for option \"-b --BitVectorComparisonMode\" are not valid.\n"</span><span class="sc">;</span>
|
|
|
755 742 <span class="s">}</span>
|
|
|
756 743 <span class="k">else</span> <span class="s">{</span>
|
|
|
757 744 <span class="k">warn</span> <span class="q">"Error: The value specified, @UnsupportedSpecifiedMeasures, for option \"-b --BitVectorComparisonMode\" is not valid.\n"</span><span class="sc">;</span>
|
|
|
758 745 <span class="s">}</span>
|
|
|
759 746 <span class="k">die</span> <span class="q">"Allowed values:"</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
760 747 <span class="s">}</span>
|
|
|
761 748 <span class="s">}</span>
|
|
|
762 749 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedComparisonMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
763 750 <span class="i">$SpecifiedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</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="sc">;</span>
|
|
|
764 751 <span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</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>
|
|
|
765 752 <span class="s">}</span>
|
|
|
766 753
|
|
|
767 754 <span class="i">$OptionsInfo</span>{<span class="w">BitVectorComparisonMode</span>} = <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>}<span class="sc">;</span>
|
|
|
768 755 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsRef</span>} = \<span class="i">@SpecifiedComparisonMeasures</span><span class="sc">;</span>
|
|
|
769 756 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsNameRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresNameMap</span><span class="sc">;</span>
|
|
|
770 757 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsMethodRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresMethodMap</span><span class="sc">;</span>
|
|
|
771 758
|
|
|
772 759 <span class="c"># Make sure valid alpha parameter is specified for Tversky calculation...</span>
|
|
|
773 760 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedMeasure1</span><span class="cm">,</span> <span class="i">$SpecifiedMeasure2</span><span class="s">)</span><span class="sc">;</span>
|
|
|
774 761 <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
775 762 <span class="i">$SpecifiedMeasure1</span> = <span class="q">'TverskySimilarity'</span><span class="sc">;</span>
|
|
|
776 763 <span class="i">$SpecifiedMeasure2</span> = <span class="q">'WeightedTverskySimilarity'</span><span class="sc">;</span>
|
|
|
777 764 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure1</span><span class="s">)</span>} || <span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure2</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
778 765 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">alpha</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
779 766 <span class="k">die</span> <span class="q">"Error: You must specify a value for \"-a, --alpha\" option in \"$SpecifiedMeasure1, $SpecifiedMeasure2, or All\" \"-m --mode\". \n"</span><span class="sc">;</span>
|
|
|
780 767 <span class="s">}</span>
|
|
|
781 768 <span class="k">my</span><span class="s">(</span><span class="i">$Alpha</span><span class="s">)</span><span class="sc">;</span>
|
|
|
782 769 <span class="i">$Alpha</span> = <span class="i">$Options</span>{<span class="w">alpha</span>}<span class="sc">;</span>
|
|
|
783 770 <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">$Alpha</span><span class="s">)</span> && <span class="i">$Alpha</span> >=<span class="n">0</span> && <span class="i">$Alpha</span> <= <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
784 771 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{alpha}, for option \"-a, --alpha\" is not valid. Allowed values: >= 0 and <= 1\n"</span><span class="sc">;</span>
|
|
|
785 772 <span class="s">}</span>
|
|
|
786 773 <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>} = <span class="i">$Alpha</span><span class="sc">;</span>
|
|
|
787 774 <span class="s">}</span>
|
|
|
788 775
|
|
|
789 776 <span class="c"># Make sure valid beta parameter is specified for WeightedTanimoto and WeightedTversky</span>
|
|
|
790 777 <span class="c"># calculations...</span>
|
|
|
791 778 <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
792 779 <span class="i">$SpecifiedMeasure1</span> = <span class="q">'WeightedTverskySimilarity'</span><span class="sc">;</span>
|
|
|
793 780 <span class="i">$SpecifiedMeasure2</span> = <span class="q">'WeightedTanimotoSimilarity'</span><span class="sc">;</span>
|
|
|
794 781 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure1</span><span class="s">)</span>} || <span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure2</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
795 782 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">beta</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
796 783 <span class="k">die</span> <span class="q">"Error: You must specify a value for \"-b, --beta\" option in \"$SpecifiedMeasure1, $SpecifiedMeasure2, or All\" \"-m --mode\". \n"</span><span class="sc">;</span>
|
|
|
797 784 <span class="s">}</span>
|
|
|
798 785 <span class="k">my</span><span class="s">(</span><span class="i">$Beta</span><span class="s">)</span><span class="sc">;</span>
|
|
|
799 786 <span class="i">$Beta</span> = <span class="i">$Options</span>{<span class="w">beta</span>}<span class="sc">;</span>
|
|
|
800 787 <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">$Beta</span><span class="s">)</span> && <span class="i">$Beta</span> >=<span class="n">0</span> && <span class="i">$Beta</span> <= <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
801 788 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{beta}, for option \"-b, --beta\" is not valid. Allowed values: >= 0 and <= 1\n"</span><span class="sc">;</span>
|
|
|
802 789 <span class="s">}</span>
|
|
|
803 790 <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>} = <span class="i">$Beta</span><span class="sc">;</span>
|
|
|
804 791 <span class="s">}</span>
|
|
|
805 792
|
|
|
806 793 <span class="c"># Setup any parameters required for specified comparison menthod...</span>
|
|
|
807 794 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedComparisonMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
808 795 <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
809 796 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedMeasure</span> =~ <span class="q">/^TverskySimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
810 797 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>}<span class="sc">;</span>
|
|
|
811 798 <span class="s">}</span>
|
|
|
812 799 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedMeasure</span> =~ <span class="q">/^WeightedTverskySimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
813 800 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Alpha</span>}<span class="sc">;</span>
|
|
|
814 801 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>}<span class="sc">;</span>
|
|
|
815 802 <span class="s">}</span>
|
|
|
816 803 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedMeasure</span> =~ <span class="q">/^WeightedTanimotoSimilarity$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
817 804 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Beta</span>}<span class="sc">;</span>
|
|
|
818 805 <span class="s">}</span>
|
|
|
819 806 <span class="s">}</span>
|
|
|
820 807 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedBitVectorComparisonsParameterRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresParameterMap</span><span class="sc">;</span>
|
|
|
821 808 <span class="s">}</span>
|
|
|
822 809
|
|
|
823 810 <span class="c"># Process options related to comparion of vector strings...</span>
|
|
|
824 811 <span class="c">#</span>
|
|
|
825 <a name="ProcessVectorComparisonOptions-"></a> 812 <span class="k">sub </span><span class="m">ProcessVectorComparisonOptions</span> <span class="s">{</span>
|
|
|
826 813 <span class="c"># Setup specified similarity coefficients for vector strings..</span>
|
|
|
827 814 <span class="k">my</span><span class="s">(</span><span class="i">$ComparisonMeasure</span><span class="cm">,</span> <span class="i">$SupportedComparisonMeasure</span><span class="cm">,</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresNameMap</span><span class="cm">,</span> <span class="i">%SupportedComparisonMeasuresMethodMap</span><span class="s">)</span><span class="sc">;</span>
|
|
|
828 815
|
|
|
829 816 <span class="i">@SupportedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
830 817 <span class="i">%SupportedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
831 818 <span class="i">%SupportedComparisonMeasuresMethodMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
832 819 <span class="k">for</span> <span class="i">$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>
|
|
|
833 820 <span class="c"># Similarity and distance coefficient function/method names contain "Coefficient" in their names.</span>
|
|
|
834 821 <span class="c"># So take 'em out and setup a map to original function/method name...</span>
|
|
|
835 822 <span class="i">$ComparisonMeasure</span> = <span class="i">$SupportedComparisonMeasure</span><span class="sc">;</span>
|
|
|
836 823 <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>
|
|
|
837 824 <span class="i">$ComparisonMeasure</span> =~ <span class="q">s/Coefficient$//i</span><span class="sc">;</span>
|
|
|
838 825 <span class="s">}</span>
|
|
|
839 826 <span class="k">push</span> <span class="i">@SupportedComparisonMeasures</span><span class="cm">,</span> <span class="i">$ComparisonMeasure</span><span class="sc">;</span>
|
|
|
840 827 <span class="i">$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>
|
|
|
841 828 <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">$SupportedComparisonMeasure</span><span class="sc">;</span>
|
|
|
842 829 <span class="s">}</span>
|
|
|
843 830
|
|
|
844 831 <span class="c"># Setup a list of similarity coefficients to use for calculating similarity matrices for bit vector strings...</span>
|
|
|
845 832 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="cm">,</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresNameMap</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresMethodMap</span><span class="cm">,</span> <span class="i">%SpecifiedComparisonMeasuresParameterMap</span><span class="s">)</span><span class="sc">;</span>
|
|
|
846 833
|
|
|
847 834 <span class="i">@SpecifiedComparisonMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
848 835 <span class="i">%SpecifiedComparisonMeasuresNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
849 836 <span class="i">%SpecifiedComparisonMeasuresMethodMap</span> = <span class="s">(</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">$Options</span>{<span class="w">vectorcomparisonmode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
852 839 <span class="k">push</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">@SupportedComparisonMeasures</span><span class="sc">;</span>
|
|
|
853 840 <span class="s">}</span>
|
|
|
854 841 <span class="k">else</span> <span class="s">{</span>
|
|
|
855 842 <span class="c"># Comma delimited list of similarity coefficients...</span>
|
|
|
856 843 <span class="k">my</span><span class="s">(</span><span class="i">$VectorComparisonMode</span><span class="cm">,</span> <span class="i">@SpecifiedMeasures</span><span class="cm">,</span> <span class="i">@UnsupportedSpecifiedMeasures</span><span class="s">)</span><span class="sc">;</span>
|
|
|
857 844
|
|
|
858 845 <span class="i">$VectorComparisonMode</span> = <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>}<span class="sc">;</span>
|
|
|
859 846 <span class="i">$VectorComparisonMode</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
|
860 847 <span class="i">@SpecifiedMeasures</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$VectorComparisonMode</span><span class="sc">;</span>
|
|
|
861 848 <span class="i">@UnsupportedSpecifiedMeasures</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
862 849
|
|
|
863 850 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
864 851 <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>
|
|
|
865 852 <span class="k">push</span> <span class="i">@SpecifiedComparisonMeasures</span><span class="cm">,</span> <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
|
|
|
866 853 <span class="s">}</span>
|
|
|
867 854 <span class="k">else</span> <span class="s">{</span>
|
|
|
868 855 <span class="k">push</span> <span class="i">@UnsupportedSpecifiedMeasures</span><span class="cm">,</span> <span class="i">$SpecifiedMeasure</span><span class="sc">;</span>
|
|
|
869 856 <span class="s">}</span>
|
|
|
870 857 <span class="s">}</span>
|
|
|
871 858 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
872 859 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedMeasures</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
|
873 860 <span class="k">warn</span> <span class="q">"Error: The values specified - "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@UnsupportedSpecifiedMeasures</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span><span class="q">" - for option \"-v --VectorComparisonMode\" are not valid.\n"</span><span class="sc">;</span>
|
|
|
874 861 <span class="s">}</span>
|
|
|
875 862 <span class="k">else</span> <span class="s">{</span>
|
|
|
876 863 <span class="k">warn</span> <span class="q">"Error: The value specified, @UnsupportedSpecifiedMeasures, for option \"-v --VectorComparisonMode\" is not valid.\n"</span><span class="sc">;</span>
|
|
|
877 864 <span class="s">}</span>
|
|
|
878 865 <span class="k">die</span> <span class="q">"Allowed values:"</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">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
|
879 866 <span class="s">}</span>
|
|
|
880 867 <span class="s">}</span>
|
|
|
881 868 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedComparisonMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
882 869 <span class="i">$SpecifiedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>} = <span class="i">$SupportedComparisonMeasuresMethodMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
883 870 <span class="i">$SpecifiedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>} = <span class="i">$SupportedComparisonMeasuresNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}<span class="sc">;</span>
|
|
|
884 871 <span class="s">}</span>
|
|
|
885 872
|
|
|
886 873 <span class="i">$OptionsInfo</span>{<span class="w">VectorComparisonMode</span>} = <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>}<span class="sc">;</span>
|
|
|
887 874 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsRef</span>} = \<span class="i">@SpecifiedComparisonMeasures</span><span class="sc">;</span>
|
|
|
888 875 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsNameRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresNameMap</span><span class="sc">;</span>
|
|
|
889 876 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsMethodRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresMethodMap</span><span class="sc">;</span>
|
|
|
890 877
|
|
|
891 878 <span class="c"># Setup specified vector comparison calculation modes...</span>
|
|
|
892 879 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedVectorComparisonModes</span><span class="s">)</span><span class="sc">;</span>
|
|
|
893 880 <span class="i">@SpecifiedVectorComparisonModes</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
894 881 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
895 882 <span class="k">push</span> <span class="i">@SpecifiedVectorComparisonModes</span><span class="cm">,</span> <span class="s">(</span><span class="q">"AlgebraicForm"</span><span class="cm">,</span> <span class="q">"BinaryForm"</span><span class="cm">,</span> <span class="q">"SetTheoreticForm"</span><span class="s">)</span><span class="sc">;</span>
|
|
|
896 883 <span class="s">}</span>
|
|
|
897 884 <span class="k">else</span> <span class="s">{</span>
|
|
|
898 885 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedFormulism</span><span class="cm">,</span> <span class="i">@SpecifiedFormulismWords</span><span class="s">)</span><span class="sc">;</span>
|
|
|
899 886
|
|
|
900 887 <span class="i">@SpecifiedFormulismWords</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>}<span class="sc">;</span>
|
|
|
901 888 <span class="k">for</span> <span class="i">$SpecifiedFormulism</span> <span class="s">(</span><span class="i">@SpecifiedFormulismWords</span><span class="s">)</span> <span class="s">{</span>
|
|
|
902 889 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedFormulism</span> !~ <span class="q">/^(AlgebraicForm|BinaryForm|SetTheoreticForm)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
903 890 <span class="k">die</span> <span class="q">"Error: The value specified, $SpecifiedFormulism, for option \"--VectorComparisonFormulism\" is not valid. Allowed values: AlgebraicForm, BinaryForm or SetTheoreticForm\n"</span><span class="sc">;</span>
|
|
|
904 891 <span class="s">}</span>
|
|
|
905 892 <span class="k">push</span> <span class="i">@SpecifiedVectorComparisonModes</span><span class="cm">,</span> <span class="i">$SpecifiedFormulism</span><span class="sc">;</span>
|
|
|
906 893 <span class="s">}</span>
|
|
|
907 894 <span class="s">}</span>
|
|
|
908 895 <span class="i">$OptionsInfo</span>{<span class="w">VectorComparisonFormulism</span>} = <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>}<span class="sc">;</span>
|
|
|
909 896 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonModesRef</span>} = \<span class="i">@SpecifiedVectorComparisonModes</span><span class="sc">;</span>
|
|
|
910 897
|
|
|
911 898 <span class="c"># Setup any parameters required for specified comparison menthod...</span>
|
|
|
912 899 <span class="k">for</span> <span class="i">$SpecifiedMeasure</span> <span class="s">(</span><span class="i">@SpecifiedComparisonMeasures</span><span class="s">)</span> <span class="s">{</span>
|
|
|
913 900 <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
914 901 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedComparisonMeasuresParameterMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedMeasure</span><span class="s">)</span>}}<span class="cm">,</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">fast</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="s">)</span><span class="sc">;</span>
|
|
|
915 902 <span class="s">}</span>
|
|
|
916 903 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedVectorComparisonsParameterRef</span>} = \<span class="i">%SpecifiedComparisonMeasuresParameterMap</span><span class="sc">;</span>
|
|
|
917 904 <span class="s">}</span>
|
|
|
918 905
|
|
|
919 906 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
|
920 <a name="SetupScriptUsage-"></a> 907 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
|
921 908
|
|
|
922 909 <span class="c"># Retrieve all the options...</span>
|
|
|
923 910 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
924 911
|
|
|
925 912 <span class="i">$Options</span>{<span class="w">alpha</span>} = <span class="n">0.5</span><span class="sc">;</span>
|
|
|
926 913 <span class="i">$Options</span>{<span class="w">beta</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
927 914
|
|
|
928 915 <span class="i">$Options</span>{<span class="w">bitvectorcomparisonmode</span>} = <span class="q">"TanimotoSimilarity"</span><span class="sc">;</span>
|
|
|
929 916
|
|
|
930 917 <span class="i">$Options</span>{<span class="w">colmode</span>} = <span class="q">'colnum'</span><span class="sc">;</span>
|
|
|
931 918
|
|
|
932 919 <span class="i">$Options</span>{<span class="w">compoundidprefix</span>} = <span class="q">'Cmpd'</span><span class="sc">;</span>
|
|
|
933 920 <span class="i">$Options</span>{<span class="w">compoundidmode</span>} = <span class="q">'LabelPrefix'</span><span class="sc">;</span>
|
|
|
934 921
|
|
|
935 922 <span class="i">$Options</span>{<span class="w">detail</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
936 923
|
|
|
937 924 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">'comma'</span><span class="sc">;</span>
|
|
|
938 925 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">'comma'</span><span class="sc">;</span>
|
|
|
939 926
|
|
|
940 927 <span class="i">$Options</span>{<span class="w">inputdatamode</span>} = <span class="q">'LoadInMemory'</span><span class="sc">;</span>
|
|
|
941 928
|
|
|
942 929 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">'AutoDetect'</span><span class="sc">;</span>
|
|
|
943 930
|
|
|
944 931 <span class="i">$Options</span>{<span class="w">outmatrixformat</span>} = <span class="q">'RowsAndColumns'</span><span class="sc">;</span>
|
|
|
945 932
|
|
|
946 933 <span class="i">$Options</span>{<span class="w">outmatrixtype</span>} = <span class="q">'FullMatrix'</span><span class="sc">;</span>
|
|
|
947 934
|
|
|
948 935 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">'yes'</span><span class="sc">;</span>
|
|
|
949 936 <span class="i">$Options</span>{<span class="w">precision</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
|
950 937
|
|
|
951 938 <span class="i">$Options</span>{<span class="w">vectorcomparisonmode</span>} = <span class="q">"TanimotoSimilarity"</span><span class="sc">;</span>
|
|
|
952 939 <span class="i">$Options</span>{<span class="w">vectorcomparisonformulism</span>} = <span class="q">"AlgebraicForm"</span><span class="sc">;</span>
|
|
|
953 940
|
|
|
954 941 <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">"alpha=f"</span><span class="cm">,</span> <span class="q">"beta=f"</span><span class="cm">,</span> <span class="q">"bitvectorcomparisonmode|b=s"</span><span class="cm">,</span> <span class="q">"colmode|c=s"</span><span class="cm">,</span> <span class="q">"compoundidcol=s"</span><span class="cm">,</span> <span class="q">"compoundidprefix=s"</span><span class="cm">,</span> <span class="q">"compoundidfield=s"</span><span class="cm">,</span> <span class="q">"compoundidmode=s"</span><span class="cm">,</span> <span class="q">"detail|d=i"</span><span class="cm">,</span> <span class="q">"fast|f"</span><span class="cm">,</span> <span class="q">"fingerprintscol=s"</span><span class="cm">,</span> <span class="q">"fingerprintsfield=s"</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"indelim=s"</span><span class="cm">,</span> <span class="q">"inputdatamode=s"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"outmatrixformat=s"</span><span class="cm">,</span> <span class="q">"outmatrixtype=s"</span><span class="cm">,</span> <span class="q">"precision|p=s"</span><span class="cm">,</span> <span class="q">"quote|q=s"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"vectorcomparisonmode|v=s"</span><span class="cm">,</span> <span class="q">"vectorcomparisonformulism=s"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
955 942 <span class="k">die</span> <span class="q">"\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
|
956 943 <span class="s">}</span>
|
|
|
957 944 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
958 945 <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>
|
|
|
959 946 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
|
960 947 <span class="s">}</span>
|
|
|
961 948 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't chdir $Options{workingdir}: $! \n"</span><span class="sc">;</span>
|
|
|
962 949 <span class="s">}</span>
|
|
|
963 950 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} !~ <span class="q">/^(ColNum|ColLabel)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
964 951 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{colmode}, for option \"-c, --ColMode\" is not valid. Allowed values: ColNum, or ColLabel\n"</span><span class="sc">;</span>
|
|
|
965 952 <span class="s">}</span>
|
|
|
966 953 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} !~ <span class="q">/^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
967 954 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{compoundidmode}, for option \"--CompoundIDMode\" is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n"</span><span class="sc">;</span>
|
|
|
968 955 <span class="s">}</span>
|
|
|
969 956 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">detail</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
970 957 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{detail}, for option \"-d, --detail\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
|
971 958 <span class="s">}</span>
|
|
|
972 959 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">inputdatamode</span>} !~ <span class="q">/^(LoadInMemory|ScanFile)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
973 960 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{inputdatamode}, for option \"--InputDataMode\" is not valid. Allowed values: LoadInMemory or ScanFile\n"</span><span class="sc">;</span>
|
|
|
974 961 <span class="s">}</span>
|
|
|
975 962 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(AutoDetect|FingerprintsBitVectorString|FingerprintsVectorString)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
976 963 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: AutoDetect, FingerprintsBitVectorString or FingerprintsVectorString \n"</span><span class="sc">;</span>
|
|
|
977 964 <span class="s">}</span>
|
|
|
978 965 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
979 966 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{indelim}, for option \"--InDelim\" is not valid. Allowed values: comma, or semicolon\n"</span><span class="sc">;</span>
|
|
|
980 967 <span class="s">}</span>
|
|
|
981 968 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
982 969 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--OutDelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
|
983 970 <span class="s">}</span>
|
|
|
984 971 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outmatrixformat</span>} !~ <span class="q">/^(RowsAndColumns|IDPairsAndValue)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
985 972 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outmatrixformat}, for option \"--OutMatrixFormat\" is not valid. Allowed values: RowsAndColumns or IDPairsAndValue\n"</span><span class="sc">;</span>
|
|
|
986 973 <span class="s">}</span>
|
|
|
987 974 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outmatrixtype</span>} !~ <span class="q">/^(FullMatrix|UpperTriangularMatrix|LowerTriangularMatrix)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
988 975 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outmatrixtype}, for option \"--OutMatrixType\" is not valid. Allowed values: FullMatrix, UpperTriangularMatrix or LowerTriangularMatrix\n"</span><span class="sc">;</span>
|
|
|
989 976 <span class="s">}</span>
|
|
|
990 977 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
991 978 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: Yes or No\n"</span><span class="sc">;</span>
|
|
|
992 979 <span class="s">}</span>
|
|
|
993 980 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">precision</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
994 981 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{precision}, for option \"--precision\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
|
995 982 <span class="s">}</span>
|
|
|
996 983 <span class="s">}</span>
|
|
|
997 984
|
|
|
998 <a name="EOF-"></a></pre>
|
|
|
999 <p> </p>
|
|
|
1000 <br />
|
|
|
1001 <center>
|
|
|
1002 <img src="../../../images/h2o2.png">
|
|
|
1003 </center>
|
|
|
1004 </body>
|
|
|
1005 </html>
|
