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

Uploaded
author deepakjadmin
date Wed, 20 Jan 2016 09:23:18 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:4816e4a8ae95
1 <html>
2 <head>
3 <title>MayaChemTools:Code:EStateIndiciesFingerprints.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: EStateIndiciesFingerprints.pl,v $</span>
17 4 <span class="c"># $Date: 2015/02/28 20:46:19 $</span>
18 5 <span class="c"># $Revision: 1.23 $</span>
19 6 <span class="c">#</span>
20 7 <span class="c"># Author: Manish Sud &lt;msud@san.rr.com&gt;</span>
21 8 <span class="c">#</span>
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
23 10 <span class="c">#</span>
24 11 <span class="c"># This file is part of MayaChemTools.</span>
25 12 <span class="c">#</span>
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
29 16 <span class="c"># later version.</span>
30 17 <span class="c">#</span>
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
34 21 <span class="c"># details.</span>
35 22 <span class="c">#</span>
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
37 24 <span class="c"># along with MayaChemTools; if not, see &lt;http://www.gnu.org/licenses/&gt; or</span>
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
40 27 <span class="c">#</span>
41 28
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">&quot;$FindBin::Bin/../lib&quot;</span><span class="sc">;</span>
44 31 <span class="k">use</span> <span class="w">Getopt::Long</span><span class="sc">;</span>
45 32 <span class="k">use</span> <span class="w">File::Basename</span><span class="sc">;</span>
46 33 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
47 34 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
48 35 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
49 36 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
50 37 <span class="k">use</span> <span class="w">SDFileUtil</span><span class="sc">;</span>
51 38 <span class="k">use</span> <span class="w">MoleculeFileIO</span><span class="sc">;</span>
52 39 <span class="k">use</span> <span class="w">FileIO::FingerprintsSDFileIO</span><span class="sc">;</span>
53 40 <span class="k">use</span> <span class="w">FileIO::FingerprintsTextFileIO</span><span class="sc">;</span>
54 41 <span class="k">use</span> <span class="w">FileIO::FingerprintsFPFileIO</span><span class="sc">;</span>
55 42 <span class="k">use</span> <span class="w">AtomTypes::EStateAtomTypes</span><span class="sc">;</span>
56 43 <span class="k">use</span> <span class="w">Fingerprints::EStateIndiciesFingerprints</span><span class="sc">;</span>
57 44
58 45 <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>
59 46
60 47 <span class="c"># Autoflush STDOUT</span>
61 48 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
62 49
63 50 <span class="c"># Starting message...</span>
64 51 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
65 52 <span class="k">print</span> <span class="q">&quot;\n$ScriptName: Starting...\n\n&quot;</span><span class="sc">;</span>
66 53 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
67 54
68 55 <span class="c"># Get the options and setup script...</span>
69 56 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
70 57 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> &lt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
71 58 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">&quot;$FindBin::Bin/$ScriptName&quot;</span><span class="s">)</span><span class="sc">;</span>
72 59 <span class="s">}</span>
73 60
74 61 <span class="k">my</span><span class="s">(</span><span class="i">@SDFilesList</span><span class="s">)</span><span class="sc">;</span>
75 62 <span class="i">@SDFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">&quot;sdf sd&quot;</span><span class="s">)</span><span class="sc">;</span>
76 63
77 64 <span class="c"># Process options...</span>
78 65 <span class="k">print</span> <span class="q">&quot;Processing options...\n&quot;</span><span class="sc">;</span>
79 66 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
80 67 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
81 68
82 69 <span class="c"># Setup information about input files...</span>
83 70 <span class="k">print</span> <span class="q">&quot;Checking input SD file(s)...\n&quot;</span><span class="sc">;</span>
84 71 <span class="k">my</span><span class="s">(</span><span class="i">%SDFilesInfo</span><span class="s">)</span><span class="sc">;</span>
85 72 <span class="i">RetrieveSDFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
86 73
87 74 <span class="c"># Process input files..</span>
88 75 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
89 76 <span class="k">if</span> <span class="s">(</span><span class="i">@SDFilesList</span> &gt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
90 77 <span class="k">print</span> <span class="q">&quot;\nProcessing SD files...\n&quot;</span><span class="sc">;</span>
91 78 <span class="s">}</span>
92 79 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span>
93 80 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
94 81 <span class="k">print</span> <span class="q">&quot;\nProcessing file $SDFilesList[$FileIndex]...\n&quot;</span><span class="sc">;</span>
95 82 <span class="i">GenerateEStateIndiciesFingerprints</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
96 83 <span class="s">}</span>
97 84 <span class="s">}</span>
98 85 <span class="k">print</span> <span class="q">&quot;\n$ScriptName:Done...\n\n&quot;</span><span class="sc">;</span>
99 86
100 87 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
101 88 <span class="i">$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>
102 89 <span class="k">print</span> <span class="q">&quot;Total time: &quot;</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
103 90
104 91 <span class="c">###############################################################################</span>
105 92
106 93 <span class="c"># Generate fingerprints for a SD file...</span>
107 94 <span class="c">#</span>
108 <a name="GenerateEStateIndiciesFingerprints-"></a> 95 <span class="k">sub </span><span class="m">GenerateEStateIndiciesFingerprints</span> <span class="s">{</span>
109 96 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
110 97 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$IgnoredCmpdCount</span><span class="cm">,</span> <span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$MoleculeFileIO</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span><span class="sc">;</span>
111 98
112 99 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span>
113 100
114 101 <span class="c"># Setup output files...</span>
115 102 <span class="c">#</span>
116 103 <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span> = <span class="i">SetupAndOpenOutputFiles</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
117 104
118 105 <span class="i">$MoleculeFileIO</span> = <span class="w">new</span> <span class="i">MoleculeFileIO</span><span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$SDFile</span><span class="s">)</span><span class="sc">;</span>
119 106 <span class="i">$MoleculeFileIO</span><span class="i">-&gt;Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
120 107
121 108 <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
122 109 <span class="i">$IgnoredCmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
123 110
124 111 <span class="j">COMPOUND:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Molecule</span> = <span class="i">$MoleculeFileIO</span><span class="i">-&gt;ReadMolecule</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
125 112 <span class="i">$CmpdCount</span>++<span class="sc">;</span>
126 113
127 114 <span class="c"># Filter compound data before calculating fingerprints...</span>
128 115 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Filter</span>}<span class="s">)</span> <span class="s">{</span>
129 116 <span class="k">if</span> <span class="s">(</span><span class="i">CheckAndFilterCompound</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
130 117 <span class="i">$IgnoredCmpdCount</span>++<span class="sc">;</span>
131 118 <span class="k">next</span> <span class="j">COMPOUND</span><span class="sc">;</span>
132 119 <span class="s">}</span>
133 120 <span class="s">}</span>
134 121
135 122 <span class="i">$EStateIndiciesFingerprints</span> = <span class="i">GenerateMoleculeFingerprints</span><span class="s">(</span><span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
136 123 <span class="k">if</span> <span class="s">(</span>!<span class="i">$EStateIndiciesFingerprints</span><span class="s">)</span> <span class="s">{</span>
137 124 <span class="i">$IgnoredCmpdCount</span>++<span class="sc">;</span>
138 125 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">&#39;FingerprintsGenerationFailed&#39;</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
139 126 <span class="k">next</span> <span class="j">COMPOUND</span><span class="sc">;</span>
140 127 <span class="s">}</span>
141 128
142 129 <span class="i">WriteDataToOutputFiles</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span><span class="sc">;</span>
143 130 <span class="s">}</span>
144 131 <span class="i">$MoleculeFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
145 132
146 133 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="s">)</span> <span class="s">{</span>
147 134 <span class="i">$NewFPSDFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
148 135 <span class="s">}</span>
149 136 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span><span class="s">)</span> <span class="s">{</span>
150 137 <span class="i">$NewFPTextFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
151 138 <span class="s">}</span>
152 139 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
153 140 <span class="i">$NewFPFileIO</span><span class="i">-&gt;Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
154 141 <span class="s">}</span>
155 142
156 143 <span class="i">WriteFingerprintsGenerationSummaryStatistics</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$IgnoredCmpdCount</span><span class="s">)</span><span class="sc">;</span>
157 144 <span class="s">}</span>
158 145
159 146 <span class="c"># Process compound being ignored due to problems in fingerprints geneation...</span>
160 147 <span class="c">#</span>
161 <a name="ProcessIgnoredCompound-"></a> 148 <span class="k">sub </span><span class="m">ProcessIgnoredCompound</span> <span class="s">{</span>
162 149 <span class="k">my</span><span class="s">(</span><span class="i">$Mode</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
163 150 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdID</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
164 151
165 152 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetDataFieldLabelAndValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
166 153 <span class="i">$CmpdID</span> = <span class="i">SetupCmpdIDForOutputFiles</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
167 154
168 155 <span class="j">MODE:</span> <span class="s">{</span>
169 156 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^ContainsNonElementalData$/i</span><span class="s">)</span> <span class="s">{</span>
170 157 <span class="k">warn</span> <span class="q">&quot;\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains atom data corresponding to non-elemental atom symbol(s)...\n\n&quot;</span><span class="sc">;</span>
171 158 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
172 159 <span class="s">}</span>
173 160
174 161 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^ContainsNoElementalData$/i</span><span class="s">)</span> <span class="s">{</span>
175 162 <span class="k">warn</span> <span class="q">&quot;\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains no atom data...\n\n&quot;</span><span class="sc">;</span>
176 163 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
177 164 <span class="s">}</span>
178 165
179 166 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^FingerprintsGenerationFailed$/i</span><span class="s">)</span> <span class="s">{</span>
180 167 <span class="k">warn</span> <span class="q">&quot;\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn&#39;t succeed...\n\n&quot;</span><span class="sc">;</span>
181 168 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
182 169 <span class="s">}</span>
183 170 <span class="k">warn</span> <span class="q">&quot;\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn&#39;t succeed...\n\n&quot;</span><span class="sc">;</span>
184 171 <span class="s">}</span>
185 172 <span class="s">}</span>
186 173
187 174 <span class="c"># Check and filter compounds....</span>
188 175 <span class="c">#</span>
189 <a name="CheckAndFilterCompound-"></a> 176 <span class="k">sub </span><span class="m">CheckAndFilterCompound</span> <span class="s">{</span>
190 177 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
191 178 <span class="k">my</span><span class="s">(</span><span class="i">$ElementCount</span><span class="cm">,</span> <span class="i">$NonElementCount</span><span class="s">)</span><span class="sc">;</span>
192 179
193 180 <span class="s">(</span><span class="i">$ElementCount</span><span class="cm">,</span> <span class="i">$NonElementCount</span><span class="s">)</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetNumOfElementsAndNonElements</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
194 181
195 182 <span class="k">if</span> <span class="s">(</span><span class="i">$NonElementCount</span><span class="s">)</span> <span class="s">{</span>
196 183 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">&#39;ContainsNonElementalData&#39;</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
197 184 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
198 185 <span class="s">}</span>
199 186
200 187 <span class="k">if</span> <span class="s">(</span>!<span class="i">$ElementCount</span><span class="s">)</span> <span class="s">{</span>
201 188 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">&#39;ContainsNoElementalData&#39;</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
202 189 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
203 190 <span class="s">}</span>
204 191
205 192 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
206 193 <span class="s">}</span>
207 194
208 195 <span class="c"># Write out compounds fingerprints generation summary statistics...</span>
209 196 <span class="c">#</span>
210 <a name="WriteFingerprintsGenerationSummaryStatistics-"></a> 197 <span class="k">sub </span><span class="m">WriteFingerprintsGenerationSummaryStatistics</span> <span class="s">{</span>
211 198 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$IgnoredCmpdCount</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
212 199 <span class="k">my</span><span class="s">(</span><span class="i">$ProcessedCmpdCount</span><span class="s">)</span><span class="sc">;</span>
213 200
214 201 <span class="i">$ProcessedCmpdCount</span> = <span class="i">$CmpdCount</span> - <span class="i">$IgnoredCmpdCount</span><span class="sc">;</span>
215 202
216 203 <span class="k">print</span> <span class="q">&quot;\nNumber of compounds: $CmpdCount\n&quot;</span><span class="sc">;</span>
217 204 <span class="k">print</span> <span class="q">&quot;Number of compounds processed successfully during fingerprints generation: $ProcessedCmpdCount\n&quot;</span><span class="sc">;</span>
218 205 <span class="k">print</span> <span class="q">&quot;Number of compounds ignored during fingerprints generation: $IgnoredCmpdCount\n&quot;</span><span class="sc">;</span>
219 206 <span class="s">}</span>
220 207
221 208 <span class="c"># Open output files...</span>
222 209 <span class="c">#</span>
223 <a name="SetupAndOpenOutputFiles-"></a> 210 <span class="k">sub </span><span class="m">SetupAndOpenOutputFiles</span> <span class="s">{</span>
224 211 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
225 212 <span class="k">my</span><span class="s">(</span><span class="i">$NewFPSDFile</span><span class="cm">,</span> <span class="i">$NewFPFile</span><span class="cm">,</span> <span class="i">$NewFPTextFile</span><span class="cm">,</span> <span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="s">)</span><span class="sc">;</span>
226 213
227 214 <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span> = <span class="s">(</span><span class="k">undef</span><span class="s">)</span> x <span class="n">3</span><span class="sc">;</span>
228 215
229 216 <span class="c"># Setup common parameters for fingerprints file IO objects...</span>
230 217 <span class="c">#</span>
231 218 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
232 219 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">&#39;Mode&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;Write&#39;</span><span class="cm">,</span> <span class="q">&#39;Overwrite&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="cm">,</span> <span class="q">&#39;FingerprintsStringMode&#39;</span> <span class="cm">=&gt;</span> <span class="q">&#39;FingerprintsVectorString&#39;</span><span class="cm">,</span> <span class="q">&#39;VectorStringFormat&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">VectorStringFormat</span>}<span class="s">)</span><span class="sc">;</span>
233 220
234 221 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
235 222 <span class="i">$NewFPSDFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
236 223 <span class="k">print</span> <span class="q">&quot;Generating SD file $NewFPSDFile...\n&quot;</span><span class="sc">;</span>
237 224 <span class="i">$NewFPSDFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsSDFileIO</span><span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$NewFPSDFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="cm">,</span> <span class="q">&#39;FingerprintsFieldLabel&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>}<span class="s">)</span><span class="sc">;</span>
238 225 <span class="i">$NewFPSDFileIO</span><span class="i">-&gt;Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
239 226 <span class="s">}</span>
240 227
241 228 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">FPOutput</span>}<span class="s">)</span> <span class="s">{</span>
242 229 <span class="i">$NewFPFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
243 230 <span class="k">print</span> <span class="q">&quot;Generating FP file $NewFPFile...\n&quot;</span><span class="sc">;</span>
244 231 <span class="i">$NewFPFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsFPFileIO</span><span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$NewFPFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="s">)</span><span class="sc">;</span>
245 232 <span class="i">$NewFPFileIO</span><span class="i">-&gt;Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
246 233 <span class="s">}</span>
247 234
248 235 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
249 236 <span class="k">my</span><span class="s">(</span><span class="i">$ColLabelsRef</span><span class="s">)</span><span class="sc">;</span>
250 237
251 238 <span class="i">$NewFPTextFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
252 239 <span class="i">$ColLabelsRef</span> = <span class="i">SetupFPTextFileCoulmnLabels</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
253 240
254 241 <span class="k">print</span> <span class="q">&quot;Generating text file $NewFPTextFile...\n&quot;</span><span class="sc">;</span>
255 242 <span class="i">$NewFPTextFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsTextFileIO</span><span class="s">(</span><span class="q">&#39;Name&#39;</span> <span class="cm">=&gt;</span> <span class="i">$NewFPTextFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="cm">,</span> <span class="q">&#39;DataColLabels&#39;</span> <span class="cm">=&gt;</span> <span class="i">$ColLabelsRef</span><span class="cm">,</span> <span class="q">&#39;OutDelim&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="q">&#39;OutQuote&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
256 243 <span class="i">$NewFPTextFileIO</span><span class="i">-&gt;Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
257 244 <span class="s">}</span>
258 245
259 246 <span class="k">return</span> <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span><span class="sc">;</span>
260 247 <span class="s">}</span>
261 248
262 249 <span class="c"># Write fingerpritns and other data to appropriate output files...</span>
263 250 <span class="c">#</span>
264 <a name="WriteDataToOutputFiles-"></a> 251 <span class="k">sub </span><span class="m">WriteDataToOutputFiles</span> <span class="s">{</span>
265 252 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$NewFPSDFileIO</span><span class="cm">,</span> <span class="i">$NewFPTextFileIO</span><span class="cm">,</span> <span class="i">$NewFPFileIO</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
266 253 <span class="k">my</span><span class="s">(</span><span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
267 254
268 255 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="k">undef</span><span class="sc">;</span>
269 256 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span> || <span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
270 257 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetDataFieldLabelAndValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
271 258 <span class="s">}</span>
272 259
273 260 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="s">)</span> <span class="s">{</span>
274 261 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdString</span><span class="s">)</span><span class="sc">;</span>
275 262
276 263 <span class="i">$CmpdString</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetInputMoleculeString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
277 264 <span class="i">$NewFPSDFileIO</span><span class="i">-&gt;WriteFingerprints</span><span class="s">(</span><span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="s">)</span><span class="sc">;</span>
278 265 <span class="s">}</span>
279 266
280 267 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span><span class="s">)</span> <span class="s">{</span>
281 268 <span class="k">my</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
282 269
283 270 <span class="i">$ColValuesRef</span> = <span class="i">SetupFPTextFileCoulmnValues</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
284 271 <span class="i">$NewFPTextFileIO</span><span class="i">-&gt;WriteFingerprints</span><span class="s">(</span><span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
285 272 <span class="s">}</span>
286 273
287 274 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
288 275 <span class="k">my</span><span class="s">(</span><span class="i">$CompoundID</span><span class="s">)</span><span class="sc">;</span>
289 276
290 277 <span class="i">$CompoundID</span> = <span class="i">SetupCmpdIDForOutputFiles</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
291 278 <span class="i">$NewFPFileIO</span><span class="i">-&gt;WriteFingerprints</span><span class="s">(</span><span class="i">$EStateIndiciesFingerprints</span><span class="cm">,</span> <span class="i">$CompoundID</span><span class="s">)</span><span class="sc">;</span>
292 279 <span class="s">}</span>
293 280
294 281 <span class="s">}</span>
295 282
296 283 <span class="c"># Generate approriate column labels for FPText output file...</span>
297 284 <span class="c">#</span>
298 <a name="SetupFPTextFileCoulmnLabels-"></a> 285 <span class="k">sub </span><span class="m">SetupFPTextFileCoulmnLabels</span> <span class="s">{</span>
299 286 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
300 287 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="s">)</span><span class="sc">;</span>
301 288
302 289 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
303 290 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
304 291 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
305 292 <span class="s">}</span>
306 293 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span> <span class="s">{</span>
307 294 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
308 295 <span class="s">}</span>
309 296 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
310 297 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="sc">;</span>
311 298 <span class="s">}</span>
312 299 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
313 300 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDLabel</span>}<span class="sc">;</span>
314 301 <span class="s">}</span>
315 302 <span class="c"># Add fingerprints label...</span>
316 303 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>}<span class="sc">;</span>
317 304
318 305 <span class="k">return</span> \<span class="i">@ColLabels</span><span class="sc">;</span>
319 306 <span class="s">}</span>
320 307
321 308 <span class="c"># Generate column values FPText output file..</span>
322 309 <span class="c">#</span>
323 <a name="SetupFPTextFileCoulmnValues-"></a> 310 <span class="k">sub </span><span class="m">SetupFPTextFileCoulmnValues</span> <span class="s">{</span>
324 311 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
325 312 <span class="k">my</span><span class="s">(</span><span class="i">@ColValues</span><span class="s">)</span><span class="sc">;</span>
326 313
327 314 <span class="i">@ColValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
328 315 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
329 316 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">SetupCmpdIDForOutputFiles</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
330 317 <span class="s">}</span>
331 318 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
332 319 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
333 320 <span class="s">}</span>
334 321 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span> <span class="s">{</span>
335 322 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
336 323 <span class="s">}</span>
337 324 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
338 325 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$_</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="sc">;</span>
339 326 <span class="s">}</span>
340 327
341 328 <span class="k">return</span> \<span class="i">@ColValues</span><span class="sc">;</span>
342 329 <span class="s">}</span>
343 330
344 331 <span class="c"># Generate compound ID for FP and FPText output files..</span>
345 332 <span class="c">#</span>
346 <a name="SetupCmpdIDForOutputFiles-"></a> 333 <span class="k">sub </span><span class="m">SetupCmpdIDForOutputFiles</span> <span class="s">{</span>
347 334 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
348 335 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdID</span><span class="s">)</span><span class="sc">;</span>
349 336
350 337 <span class="i">$CmpdID</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
351 338 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^MolNameOrLabelPrefix$/i</span><span class="s">)</span> <span class="s">{</span>
352 339 <span class="k">my</span><span class="s">(</span><span class="i">$MolName</span><span class="s">)</span><span class="sc">;</span>
353 340 <span class="i">$MolName</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetName</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
354 341 <span class="i">$CmpdID</span> = <span class="i">$MolName</span> ? <span class="i">$MolName</span> <span class="co">:</span> <span class="q">&quot;$OptionsInfo{CompoundID}${CmpdCount}&quot;</span><span class="sc">;</span>
355 342 <span class="s">}</span>
356 343 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^LabelPrefix$/i</span><span class="s">)</span> <span class="s">{</span>
357 344 <span class="i">$CmpdID</span> = <span class="q">&quot;$OptionsInfo{CompoundID}${CmpdCount}&quot;</span><span class="sc">;</span>
358 345 <span class="s">}</span>
359 346 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span>
360 347 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedDataField</span><span class="s">)</span><span class="sc">;</span>
361 348 <span class="i">$SpecifiedDataField</span> = <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>}<span class="sc">;</span>
362 349 <span class="i">$CmpdID</span> = <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$SpecifiedDataField</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>-&gt;{<span class="i">$SpecifiedDataField</span>} <span class="co">:</span> <span class="q">&#39;&#39;</span><span class="sc">;</span>
363 350 <span class="s">}</span>
364 351 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^MolName$/i</span><span class="s">)</span> <span class="s">{</span>
365 352 <span class="i">$CmpdID</span> = <span class="i">$Molecule</span><span class="i">-&gt;GetName</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
366 353 <span class="s">}</span>
367 354 <span class="k">return</span> <span class="i">$CmpdID</span><span class="sc">;</span>
368 355 <span class="s">}</span>
369 356
370 357 <span class="c"># Generate fingerprints for molecule...</span>
371 358 <span class="c">#</span>
372 <a name="GenerateMoleculeFingerprints-"></a> 359 <span class="k">sub </span><span class="m">GenerateMoleculeFingerprints</span> <span class="s">{</span>
373 360 <span class="k">my</span><span class="s">(</span><span class="i">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
374 361 <span class="k">my</span><span class="s">(</span><span class="i">$EStateIndiciesFingerprints</span><span class="s">)</span><span class="sc">;</span>
375 362
376 363 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">KeepLargestComponent</span>}<span class="s">)</span> <span class="s">{</span>
377 364 <span class="i">$Molecule</span><span class="i">-&gt;KeepLargestComponent</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
378 365 <span class="s">}</span>
379 366 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Molecule</span><span class="i">-&gt;DetectRings</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
380 367 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
381 368 <span class="s">}</span>
382 369 <span class="i">$Molecule</span><span class="i">-&gt;SetAromaticityModel</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AromaticityModel</span>}<span class="s">)</span><span class="sc">;</span>
383 370 <span class="i">$Molecule</span><span class="i">-&gt;DetectAromaticity</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
384 371
385 372 <span class="i">$EStateIndiciesFingerprints</span> = <span class="w">new</span> <span class="i">Fingerprints::EStateIndiciesFingerprints</span><span class="s">(</span><span class="q">&#39;Molecule&#39;</span> <span class="cm">=&gt;</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="q">&#39;EStateAtomTypesSetToUse&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">EStateAtomTypesSetToUse</span>}<span class="cm">,</span> <span class="q">&#39;ValuesPrecision&#39;</span> <span class="cm">=&gt;</span> <span class="i">$OptionsInfo</span>{<span class="w">ValuesPrecision</span>}<span class="s">)</span><span class="sc">;</span>
386 373
387 374 <span class="c"># Generate E-state indicies fingerprints...</span>
388 375 <span class="i">$EStateIndiciesFingerprints</span><span class="i">-&gt;GenerateFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
389 376
390 377 <span class="c"># Make sure E-state indicies fingerprints generation is successful...</span>
391 378 <span class="k">if</span> <span class="s">(</span>!<span class="i">$EStateIndiciesFingerprints</span><span class="i">-&gt;IsFingerprintsGenerationSuccessful</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
392 379 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
393 380 <span class="s">}</span>
394 381
395 382 <span class="k">return</span> <span class="i">$EStateIndiciesFingerprints</span><span class="sc">;</span>
396 383 <span class="s">}</span>
397 384
398 385 <span class="c"># Retrieve information about SD files...</span>
399 386 <span class="c">#</span>
400 <a name="RetrieveSDFilesInfo-"></a> 387 <span class="k">sub </span><span class="m">RetrieveSDFilesInfo</span> <span class="s">{</span>
401 388 <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$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">$FileName</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$TextOutFileExt</span><span class="cm">,</span> <span class="i">$SDOutFileExt</span><span class="cm">,</span> <span class="i">$FPOutFileExt</span><span class="cm">,</span> <span class="i">$NewSDFileName</span><span class="cm">,</span> <span class="i">$NewFPFileName</span><span class="cm">,</span> <span class="i">$NewTextFileName</span><span class="cm">,</span> <span class="i">$CheckDataField</span><span class="cm">,</span> <span class="i">$CollectDataFields</span><span class="cm">,</span> <span class="i">$AllDataFieldsRef</span><span class="cm">,</span> <span class="i">$CommonDataFieldsRef</span><span class="s">)</span><span class="sc">;</span>
402 389
403 390 <span class="i">%SDFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
404 391 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
405 392 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
406 393 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
407 394 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
408 395 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
409 396 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
410 397 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
411 398
412 399 <span class="i">$CheckDataField</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} &amp;&amp; <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> &amp;&amp; <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
413 400 <span class="i">$CollectDataFields</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} &amp;&amp; <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^(All|Common)$/i</span><span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
414 401
415 402 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span>
416 403 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
417 404
418 405 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
419 406 <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">&#39;&#39;</span><span class="sc">;</span>
420 407 <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">&#39;&#39;</span><span class="sc">;</span>
421 408 <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">&#39;&#39;</span><span class="sc">;</span>
422 409 <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">&#39;&#39;</span><span class="sc">;</span>
423 410
424 411 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
425 412 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$SDFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
426 413 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: It doesn&#39;t exist\n&quot;</span><span class="sc">;</span>
427 414 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
428 415 <span class="s">}</span>
429 416 <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="q">&quot;sd sdf&quot;</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
430 417 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: It&#39;s not a SD file\n&quot;</span><span class="sc">;</span>
431 418 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
432 419 <span class="s">}</span>
433 420
434 421 <span class="k">if</span> <span class="s">(</span><span class="i">$CheckDataField</span><span class="s">)</span> <span class="s">{</span>
435 422 <span class="c"># Make sure data field exists in SD file..</span>
436 423 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdString</span><span class="cm">,</span> <span class="i">$SpecifiedDataField</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">%DataFieldValues</span><span class="s">)</span><span class="sc">;</span>
437 424
438 425 <span class="i">@CmpdLines</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
439 426 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">&quot;$SDFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $SDFile: $! \n&quot;</span><span class="sc">;</span>
440 427 <span class="i">$CmpdString</span> = <span class="i">ReadCmpdString</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="sc">;</span>
441 428 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
442 429 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">&quot;\n&quot;</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span>
443 430 <span class="i">%DataFieldValues</span> = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
444 431 <span class="i">$SpecifiedDataField</span> = <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>}<span class="sc">;</span>
445 432 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$DataFieldValues</span>{<span class="i">$SpecifiedDataField</span>}<span class="s">)</span> <span class="s">{</span>
446 433 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: Data field value, $SpecifiedDataField, using \&quot;--CompoundID\&quot; option in \&quot;DataField\&quot; \&quot;--CompoundIDMode\&quot; doesn&#39;t exist\n&quot;</span><span class="sc">;</span>
447 434 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
448 435 <span class="s">}</span>
449 436 <span class="s">}</span>
450 437
451 438 <span class="i">$AllDataFieldsRef</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
452 439 <span class="i">$CommonDataFieldsRef</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
453 440 <span class="k">if</span> <span class="s">(</span><span class="i">$CollectDataFields</span><span class="s">)</span> <span class="s">{</span>
454 441 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="s">)</span><span class="sc">;</span>
455 442 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">&quot;$SDFile&quot;</span> <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t open $SDFile: $! \n&quot;</span><span class="sc">;</span>
456 443 <span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$AllDataFieldsRef</span><span class="cm">,</span> <span class="i">$CommonDataFieldsRef</span><span class="s">)</span> = <span class="i">GetAllAndCommonCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="sc">;</span>
457 444 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
458 445 <span class="s">}</span>
459 446
460 447 <span class="c"># Setup output file names...</span>
461 448 <span class="i">$FileDir</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">&quot;&quot;</span><span class="sc">;</span>
462 449 <span class="s">(</span><span class="i">$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">$SDFile</span><span class="s">)</span><span class="sc">;</span>
463 450
464 451 <span class="i">$TextOutFileExt</span> = <span class="q">&quot;csv&quot;</span><span class="sc">;</span>
465 452 <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>
466 453 <span class="i">$TextOutFileExt</span> = <span class="q">&quot;tsv&quot;</span><span class="sc">;</span>
467 454 <span class="s">}</span>
468 455 <span class="i">$SDOutFileExt</span> = <span class="i">$FileExt</span><span class="sc">;</span>
469 456 <span class="i">$FPOutFileExt</span> = <span class="q">&quot;fpf&quot;</span><span class="sc">;</span>
470 457
471 458 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} &amp;&amp; <span class="s">(</span><span class="i">@SDFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
472 459 <span class="k">my</span> <span class="s">(</span><span class="i">$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">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span>
473 460 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> &amp;&amp; <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
474 461 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
475 462 <span class="s">}</span>
476 463 <span class="k">else</span> <span class="s">{</span>
477 464 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
478 465 <span class="s">}</span>
479 466 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
480 467 <span class="s">}</span>
481 468 <span class="k">else</span> <span class="s">{</span>
482 469 <span class="i">$OutFileRoot</span> = <span class="q">&quot;${FileName}EStateIndiciesFP&quot;</span><span class="sc">;</span>
483 470 <span class="s">}</span>
484 471
485 472 <span class="i">$NewSDFileName</span> = <span class="q">&quot;${OutFileRoot}.${SDOutFileExt}&quot;</span><span class="sc">;</span>
486 473 <span class="i">$NewFPFileName</span> = <span class="q">&quot;${OutFileRoot}.${FPOutFileExt}&quot;</span><span class="sc">;</span>
487 474 <span class="i">$NewTextFileName</span> = <span class="q">&quot;${OutFileRoot}.${TextOutFileExt}&quot;</span><span class="sc">;</span>
488 475
489 476 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
490 477 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFile</span> =~ <span class="q">/$NewSDFileName/i</span><span class="s">)</span> <span class="s">{</span>
491 478 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n&quot;</span><span class="sc">;</span>
492 479 <span class="k">print</span> <span class="q">&quot;Specify a different name using \&quot;-r --root\&quot; option or use default name.\n&quot;</span><span class="sc">;</span>
493 480 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
494 481 <span class="s">}</span>
495 482 <span class="s">}</span>
496 483
497 484 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span>
498 485 <span class="c"># Check SD and text outout files...</span>
499 486 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
500 487 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewSDFileName</span><span class="s">)</span> <span class="s">{</span>
501 488 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: The file $NewSDFileName already exists\n&quot;</span><span class="sc">;</span>
502 489 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
503 490 <span class="s">}</span>
504 491 <span class="s">}</span>
505 492 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">FPOutput</span>}<span class="s">)</span> <span class="s">{</span>
506 493 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewFPFileName</span><span class="s">)</span> <span class="s">{</span>
507 494 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: The file $NewFPFileName already exists\n&quot;</span><span class="sc">;</span>
508 495 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
509 496 <span class="s">}</span>
510 497 <span class="s">}</span>
511 498 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
512 499 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewTextFileName</span><span class="s">)</span> <span class="s">{</span>
513 500 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SDFile: The file $NewTextFileName already exists\n&quot;</span><span class="sc">;</span>
514 501 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
515 502 <span class="s">}</span>
516 503 <span class="s">}</span>
517 504 <span class="s">}</span>
518 505
519 506 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
520 507
521 508 <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span>
522 509 <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewSDFileName</span><span class="sc">;</span>
523 510 <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewFPFileName</span><span class="sc">;</span>
524 511 <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewTextFileName</span><span class="sc">;</span>
525 512
526 513 <span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$Index</span>] = <span class="i">$AllDataFieldsRef</span><span class="sc">;</span>
527 514 <span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$Index</span>] = <span class="i">$CommonDataFieldsRef</span><span class="sc">;</span>
528 515 <span class="s">}</span>
529 516 <span class="s">}</span>
530 517
531 518 <span class="c"># Process option values...</span>
532 <a name="ProcessOptions-"></a> 519 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
533 520 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
534 521
535 522 <span class="i">$OptionsInfo</span>{<span class="w">AromaticityModel</span>} = <span class="i">$Options</span>{<span class="w">aromaticitymodel</span>}<span class="sc">;</span>
536 523
537 524 <span class="i">$OptionsInfo</span>{<span class="w">EStateAtomTypesSetToUse</span>} = <span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} ? <span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} <span class="co">:</span> <span class="q">&#39;ArbitrarySize&#39;</span><span class="sc">;</span>
538 525
539 526 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} = <span class="i">$Options</span>{<span class="w">compoundidmode</span>}<span class="sc">;</span>
540 527 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDLabel</span>} = <span class="i">$Options</span>{<span class="w">compoundidlabel</span>}<span class="sc">;</span>
541 528 <span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} = <span class="i">$Options</span>{<span class="w">datafieldsmode</span>}<span class="sc">;</span>
542 529
543 530 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedDataFields</span><span class="s">)</span><span class="sc">;</span>
544 531 <span class="i">@SpecifiedDataFields</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
545 532
546 533 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
547 534 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
548 535
549 536 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
550 537 <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>
551 538 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">compoundid</span>}<span class="s">)</span> <span class="s">{</span>
552 539 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--CompoundID\&quot; option in \&quot;DataField\&quot; \&quot;--CompoundIDMode\&quot;. \n&quot;</span><span class="sc">;</span>
553 540 <span class="s">}</span>
554 541 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="i">$Options</span>{<span class="w">compoundid</span>}<span class="sc">;</span>
555 542 <span class="s">}</span>
556 543 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} =~ <span class="q">/^(LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
557 544 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="i">$Options</span>{<span class="w">compoundid</span>} ? <span class="i">$Options</span>{<span class="w">compoundid</span>} <span class="co">:</span> <span class="q">&#39;Cmpd&#39;</span><span class="sc">;</span>
558 545 <span class="s">}</span>
559 546 <span class="s">}</span>
560 547 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
561 548 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
562 549 <span class="k">die</span> <span class="q">&quot;Error: You must specify a value for \&quot;--DataFields\&quot; option in \&quot;Specify\&quot; \&quot;-d, --DataFieldsMode\&quot;. \n&quot;</span><span class="sc">;</span>
563 550 <span class="s">}</span>
564 551 <span class="i">@SpecifiedDataFields</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="sc">;</span>
565 552 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="cm">,</span> <span class="i">@SpecifiedDataFields</span><span class="sc">;</span>
566 553 <span class="s">}</span>
567 554
568 555 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>} = <span class="i">$Options</span>{<span class="w">fingerprintslabel</span>} ? <span class="i">$Options</span>{<span class="w">fingerprintslabel</span>} <span class="co">:</span> <span class="q">&#39;EStateIndiciesFingerprints&#39;</span><span class="sc">;</span>
569 556
570 557 <span class="i">$OptionsInfo</span>{<span class="w">Filter</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">filter</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
571 558
572 559 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintslabelmode</span>} =~ <span class="q">/^FingerprintsLabelWithIDs$/</span><span class="s">)</span> <span class="s">{</span>
573 560 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} =~ <span class="q">/^FixedSize$/i</span><span class="s">)</span> <span class="s">{</span>
574 561 <span class="c"># Append E-state atom types for non-hydrogen atoms to the fingerprints label...</span>
575 562 <span class="k">my</span><span class="s">(</span><span class="i">$AtomType</span><span class="cm">,</span> <span class="i">@IDs</span><span class="s">)</span><span class="sc">;</span>
576 563 <span class="i">@IDs</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
577 564 <span class="k">for</span> <span class="i">$AtomType</span> <span class="s">(</span><span class="i">@</span>{<span class="i">AtomTypes::EStateAtomTypes::GetAllPossibleEStateNonHydrogenAtomTypes</span><span class="s">(</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span>
578 565 <span class="k">push</span> <span class="i">@IDs</span><span class="cm">,</span> <span class="q">&quot;S${AtomType}&quot;</span><span class="sc">;</span>
579 566 <span class="s">}</span>
580 567 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>} .= <span class="q">&quot;; EStateAtomTypes: &quot;</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@IDs</span><span class="cm">,</span> <span class="q">&quot; &quot;</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="sc">;</span>
581 568 <span class="s">}</span>
582 569 <span class="s">}</span>
583 570 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabelMode</span>} = <span class="i">$Options</span>{<span class="w">fingerprintslabelmode</span>}<span class="sc">;</span>
584 571
585 572 <span class="i">$OptionsInfo</span>{<span class="w">KeepLargestComponent</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">keeplargestcomponent</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
586 573
587 574 <span class="i">$OptionsInfo</span>{<span class="w">Output</span>} = <span class="i">$Options</span>{<span class="w">output</span>}<span class="sc">;</span>
588 575 <span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(SD|All)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
589 576 <span class="i">$OptionsInfo</span>{<span class="w">FPOutput</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(FP|All)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
590 577 <span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(Text|All)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
591 578
592 579 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="i">$Options</span>{<span class="w">outdelim</span>}<span class="sc">;</span>
593 580 <span class="i">$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>
594 581
595 582 <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>
596 583 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
597 584
598 585 <span class="c"># Precision for E-state indicies...</span>
599 586 <span class="i">$OptionsInfo</span>{<span class="w">ValuesPrecision</span>} = <span class="i">$Options</span>{<span class="w">valuesprecision</span>}<span class="sc">;</span>
600 587
601 588 <span class="c"># Setup default vector string format...</span>
602 589 <span class="k">my</span><span class="s">(</span><span class="i">$VectorStringFormat</span><span class="s">)</span><span class="sc">;</span>
603 590 <span class="i">$VectorStringFormat</span> = <span class="q">&#39;&#39;</span><span class="sc">;</span>
604 591 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">vectorstringformat</span>}<span class="s">)</span> <span class="s">{</span>
605 592 <span class="i">$VectorStringFormat</span> = <span class="i">$Options</span>{<span class="w">vectorstringformat</span>}<span class="sc">;</span>
606 593 <span class="s">}</span>
607 594 <span class="k">else</span> <span class="s">{</span>
608 595 <span class="i">$VectorStringFormat</span> = <span class="s">(</span><span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} =~ <span class="q">/^FixedSize$/</span><span class="s">)</span> ? <span class="q">&quot;ValuesString&quot;</span> <span class="co">:</span> <span class="q">&quot;IDsAndValuesString&quot;</span><span class="sc">;</span>
609 596 <span class="s">}</span>
610 597 <span class="i">$OptionsInfo</span>{<span class="w">VectorStringFormat</span>} = <span class="i">$VectorStringFormat</span><span class="sc">;</span>
611 598 <span class="s">}</span>
612 599
613 600 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
614 <a name="SetupScriptUsage-"></a> 601 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
615 602
616 603 <span class="c"># Retrieve all the options...</span>
617 604 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
618 605
619 606 <span class="i">$Options</span>{<span class="w">aromaticitymodel</span>} = <span class="q">&#39;MayaChemToolsAromaticityModel&#39;</span><span class="sc">;</span>
620 607
621 608 <span class="i">$Options</span>{<span class="w">compoundidmode</span>} = <span class="q">&#39;LabelPrefix&#39;</span><span class="sc">;</span>
622 609 <span class="i">$Options</span>{<span class="w">compoundidlabel</span>} = <span class="q">&#39;CompoundID&#39;</span><span class="sc">;</span>
623 610 <span class="i">$Options</span>{<span class="w">datafieldsmode</span>} = <span class="q">&#39;CompoundID&#39;</span><span class="sc">;</span>
624 611
625 612 <span class="i">$Options</span>{<span class="w">filter</span>} = <span class="q">&#39;Yes&#39;</span><span class="sc">;</span>
626 613
627 614 <span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} = <span class="q">&#39;ArbitrarySize&#39;</span><span class="sc">;</span>
628 615
629 616 <span class="i">$Options</span>{<span class="w">fingerprintslabelmode</span>} = <span class="q">&#39;FingerprintsLabelOnly&#39;</span><span class="sc">;</span>
630 617 <span class="i">$Options</span>{<span class="w">keeplargestcomponent</span>} = <span class="q">&#39;Yes&#39;</span><span class="sc">;</span>
631 618
632 619 <span class="i">$Options</span>{<span class="w">output</span>} = <span class="q">&#39;text&#39;</span><span class="sc">;</span>
633 620 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">&#39;comma&#39;</span><span class="sc">;</span>
634 621 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">&#39;yes&#39;</span><span class="sc">;</span>
635 622
636 623 <span class="i">$Options</span>{<span class="w">valuesprecision</span>} = <span class="n">3</span><span class="sc">;</span>
637 624
638 625 <span class="i">$Options</span>{<span class="w">vectorstringformat</span>} = <span class="q">&#39;&#39;</span><span class="sc">;</span>
639 626
640 627 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">&quot;aromaticitymodel=s&quot;</span><span class="cm">,</span> <span class="q">&quot;compoundid=s&quot;</span><span class="cm">,</span> <span class="q">&quot;compoundidlabel=s&quot;</span><span class="cm">,</span> <span class="q">&quot;compoundidmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;datafields=s&quot;</span><span class="cm">,</span> <span class="q">&quot;datafieldsmode|d=s&quot;</span><span class="cm">,</span> <span class="q">&quot;estateatomtypessettouse|e=s&quot;</span><span class="cm">,</span> <span class="q">&quot;filter|f=s&quot;</span><span class="cm">,</span> <span class="q">&quot;fingerprintslabelmode=s&quot;</span><span class="cm">,</span> <span class="q">&quot;fingerprintslabel=s&quot;</span><span class="cm">,</span> <span class="q">&quot;help|h&quot;</span><span class="cm">,</span> <span class="q">&quot;keeplargestcomponent|k=s&quot;</span><span class="cm">,</span> <span class="q">&quot;outdelim=s&quot;</span><span class="cm">,</span> <span class="q">&quot;output=s&quot;</span><span class="cm">,</span> <span class="q">&quot;overwrite|o&quot;</span><span class="cm">,</span> <span class="q">&quot;quote|q=s&quot;</span><span class="cm">,</span> <span class="q">&quot;root|r=s&quot;</span><span class="cm">,</span> <span class="q">&quot;valuesprecision=s&quot;</span><span class="cm">,</span> <span class="q">&quot;vectorstringformat|v=s&quot;</span><span class="cm">,</span> <span class="q">&quot;workingdir|w=s&quot;</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
641 628 <span class="k">die</span> <span class="q">&quot;\nTo get a list of valid options and their values, use \&quot;$ScriptName -h\&quot; or\n\&quot;perl -S $ScriptName -h\&quot; command and try again...\n&quot;</span><span class="sc">;</span>
642 629 <span class="s">}</span>
643 630 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
644 631 <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>
645 632 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{workingdir}, for option \&quot;-w --workingdir\&quot; is not a directory name.\n&quot;</span><span class="sc">;</span>
646 633 <span class="s">}</span>
647 634 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t chdir $Options{workingdir}: $! \n&quot;</span><span class="sc">;</span>
648 635 <span class="s">}</span>
649 636 <span class="k">if</span> <span class="s">(</span>!<span class="i">Molecule::IsSupportedAromaticityModel</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">aromaticitymodel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
650 637 <span class="k">my</span><span class="s">(</span><span class="i">@SupportedModels</span><span class="s">)</span> = <span class="i">Molecule::GetSupportedAromaticityModels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
651 638 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{aromaticitymodel}, for option \&quot;--AromaticityModel\&quot; is not valid. Supported aromaticity models in current release of MayaChemTools: @SupportedModels\n&quot;</span><span class="sc">;</span>
652 639 <span class="s">}</span>
653 640 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} !~ <span class="q">/^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
654 641 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{compoundidmode}, for option \&quot;--CompoundIDMode\&quot; is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n&quot;</span><span class="sc">;</span>
655 642 <span class="s">}</span>
656 643 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} !~ <span class="q">/^(All|Common|Specify|CompoundID)$/i</span><span class="s">)</span> <span class="s">{</span>
657 644 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{datafieldsmode}, for option \&quot;-d, --DataFieldsMode\&quot; is not valid. Allowed values: All, Common, Specify or CompoundID\n&quot;</span><span class="sc">;</span>
658 645 <span class="s">}</span>
659 646 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} &amp;&amp; <span class="i">$Options</span>{<span class="w">estateatomtypessettouse</span>} !~ <span class="q">/^(ArbitrarySize|FixedSize)$/</span><span class="s">)</span> <span class="s">{</span>
660 647 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{estateatomtypessettouse}, for option \&quot;-e, --EStateAtomTypesSetToUse\&quot; is not valid. Allowed values: ArbitrarySize or FixedSize\n&quot;</span><span class="sc">;</span>
661 648 <span class="s">}</span>
662 649 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">filter</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
663 650 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{filter}, for option \&quot;-f, --Filter\&quot; is not valid. Allowed values: Yes or No\n&quot;</span><span class="sc">;</span>
664 651 <span class="s">}</span>
665 652 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">fingerprintslabelmode</span>} !~ <span class="q">/^(FingerprintsLabelOnly|FingerprintsLabelWithIDs)$/i</span><span class="s">)</span> <span class="s">{</span>
666 653 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{fingerprintslabelmode}, for option \&quot;--FingerprintsLabelMode\&quot; is not valid. Allowed values: FingerprintsLabelOnly or FingerprintsLabelWithIDs\n&quot;</span><span class="sc">;</span>
667 654 <span class="s">}</span>
668 655 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">keeplargestcomponent</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
669 656 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{keeplargestcomponent}, for option \&quot;-k, --KeepLargestComponent\&quot; is not valid. Allowed values: Yes or No\n&quot;</span><span class="sc">;</span>
670 657 <span class="s">}</span>
671 658 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} !~ <span class="q">/^(SD|FP|text|all)$/i</span><span class="s">)</span> <span class="s">{</span>
672 659 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{output}, for option \&quot;--output\&quot; is not valid. Allowed values: SD, FP, text, or all\n&quot;</span><span class="sc">;</span>
673 660 <span class="s">}</span>
674 661 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
675 662 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{outdelim}, for option \&quot;--outdelim\&quot; is not valid. Allowed values: comma, tab, or semicolon\n&quot;</span><span class="sc">;</span>
676 663 <span class="s">}</span>
677 664 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
678 665 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{quote}, for option \&quot;-q --quote\&quot; is not valid. Allowed values: Yes or No\n&quot;</span><span class="sc">;</span>
679 666 <span class="s">}</span>
680 667 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span> &amp;&amp; <span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/^No$/i</span><span class="s">)</span> <span class="s">{</span>
681 668 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{quote}, for option \&quot;-q --quote\&quot; is not allowed with, semicolon value of \&quot;--outdelim\&quot; option: Fingerprints string use semicolon as delimiter for various data fields and must be quoted.\n&quot;</span><span class="sc">;</span>
682 669 <span class="s">}</span>
683 670 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">valuesprecision</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
684 671 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{valuesprecision}, for option \&quot;--ValuesPrecision\&quot; is not valid. Allowed values: &gt; 0 \n&quot;</span><span class="sc">;</span>
685 672 <span class="s">}</span>
686 673 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">vectorstringformat</span>} &amp;&amp; <span class="i">$Options</span>{<span class="w">vectorstringformat</span>} !~ <span class="q">/^(ValuesString|IDsAndValuesString|IDsAndValuesPairsString|ValuesAndIDsString|ValuesAndIDsPairsString)$/i</span><span class="s">)</span> <span class="s">{</span>
687 674 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{vectorstringformat}, for option \&quot;-v, --VectorStringFormat\&quot; is not valid. Allowed values: ValuesString, IDsAndValuesString, IDsAndValuesPairsString, ValuesAndIDsString or ValuesAndIDsPairsString\n&quot;</span><span class="sc">;</span>
688 675 <span class="s">}</span>
689 676 <span class="s">}</span>
690 677
691 <a name="EOF-"></a></pre>
692 <p>&nbsp;</p>
693 <br />
694 <center>
695 <img src="../../../images/h2o2.png">
696 </center>
697 </body>
698 </html>