0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:ExtendedConnectivityFingerprints.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: ExtendedConnectivityFingerprints.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:19 $</span>
|
|
18 5 <span class="c"># $Revision: 1.37 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$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">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::AtomicInvariantsAtomTypes</span><span class="sc">;</span>
|
|
56 43 <span class="k">use</span> <span class="w">AtomTypes::FunctionalClassAtomTypes</span><span class="sc">;</span>
|
|
57 44 <span class="k">use</span> <span class="w">Fingerprints::ExtendedConnectivityFingerprints</span><span class="sc">;</span>
|
|
58 45
|
|
59 46 <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>
|
|
60 47
|
|
61 48 <span class="c"># Autoflush STDOUT</span>
|
|
62 49 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
63 50
|
|
64 51 <span class="c"># Starting message...</span>
|
|
65 52 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
66 53 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span>
|
|
67 54 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
68 55
|
|
69 56 <span class="c"># Get the options and setup script...</span>
|
|
70 57 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
71 58 <span class="k">if</span> <span class="s">(</span><span class="i">$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>
|
|
72 59 <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>
|
|
73 60 <span class="s">}</span>
|
|
74 61
|
|
75 62 <span class="k">my</span><span class="s">(</span><span class="i">@SDFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
76 63 <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">"sdf sd"</span><span class="s">)</span><span class="sc">;</span>
|
|
77 64
|
|
78 65 <span class="c"># Process options...</span>
|
|
79 66 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
80 67 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
81 68 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
82 69
|
|
83 70 <span class="c"># Setup information about input files...</span>
|
|
84 71 <span class="k">print</span> <span class="q">"Checking input SD file(s)...\n"</span><span class="sc">;</span>
|
|
85 72 <span class="k">my</span><span class="s">(</span><span class="i">%SDFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
86 73 <span class="i">RetrieveSDFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
87 74
|
|
88 75 <span class="c"># Process input files..</span>
|
|
89 76 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
90 77 <span class="k">if</span> <span class="s">(</span><span class="i">@SDFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
91 78 <span class="k">print</span> <span class="q">"\nProcessing SD files...\n"</span><span class="sc">;</span>
|
|
92 79 <span class="s">}</span>
|
|
93 80 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
94 81 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
95 82 <span class="k">print</span> <span class="q">"\nProcessing file $SDFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
96 83 <span class="i">GenerateExtendedConnectivityFingerprints</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
97 84 <span class="s">}</span>
|
|
98 85 <span class="s">}</span>
|
|
99 86 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
100 87
|
|
101 88 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
102 89 <span class="i">$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>
|
|
103 90 <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">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
104 91
|
|
105 92 <span class="c">###############################################################################</span>
|
|
106 93
|
|
107 94 <span class="c"># Generate fingerprints for a SD file...</span>
|
|
108 95 <span class="c">#</span>
|
|
109 <a name="GenerateExtendedConnectivityFingerprints-"></a> 96 <span class="k">sub </span><span class="m">GenerateExtendedConnectivityFingerprints</span> <span class="s">{</span>
|
|
110 97 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
111 98 <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">$Molecule</span><span class="cm">,</span> <span class="i">$ExtendedConnectivityFingerprints</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">$NewFPFileIO</span><span class="s">)</span><span class="sc">;</span>
|
|
112 99
|
|
113 100 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
114 101
|
|
115 102 <span class="c"># Setup output files...</span>
|
|
116 103 <span class="c">#</span>
|
|
117 104 <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>
|
|
118 105
|
|
119 106 <span class="i">$MoleculeFileIO</span> = <span class="w">new</span> <span class="i">MoleculeFileIO</span><span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$SDFile</span><span class="s">)</span><span class="sc">;</span>
|
|
120 107 <span class="i">$MoleculeFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
121 108
|
|
122 109 <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
123 110 <span class="i">$IgnoredCmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
124 111
|
|
125 112 <span class="j">COMPOUND:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Molecule</span> = <span class="i">$MoleculeFileIO</span><span class="i">->ReadMolecule</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
126 113 <span class="i">$CmpdCount</span>++<span class="sc">;</span>
|
|
127 114
|
|
128 115 <span class="c"># Filter compound data before calculating fingerprints...</span>
|
|
129 116 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Filter</span>}<span class="s">)</span> <span class="s">{</span>
|
|
130 117 <span class="k">if</span> <span class="s">(</span><span class="i">CheckAndFilterCompound</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
131 118 <span class="i">$IgnoredCmpdCount</span>++<span class="sc">;</span>
|
|
132 119 <span class="k">next</span> <span class="j">COMPOUND</span><span class="sc">;</span>
|
|
133 120 <span class="s">}</span>
|
|
134 121 <span class="s">}</span>
|
|
135 122
|
|
136 123 <span class="i">$ExtendedConnectivityFingerprints</span> = <span class="i">GenerateMoleculeFingerprints</span><span class="s">(</span><span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
|
|
137 124 <span class="k">if</span> <span class="s">(</span>!<span class="i">$ExtendedConnectivityFingerprints</span><span class="s">)</span> <span class="s">{</span>
|
|
138 125 <span class="i">$IgnoredCmpdCount</span>++<span class="sc">;</span>
|
|
139 126 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">'FingerprintsGenerationFailed'</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>
|
|
140 127 <span class="k">next</span> <span class="j">COMPOUND</span><span class="sc">;</span>
|
|
141 128 <span class="s">}</span>
|
|
142 129
|
|
143 130 <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">$Molecule</span><span class="cm">,</span> <span class="i">$ExtendedConnectivityFingerprints</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">$NewFPFileIO</span><span class="s">)</span><span class="sc">;</span>
|
|
144 131 <span class="s">}</span>
|
|
145 132 <span class="i">$MoleculeFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
146 133
|
|
147 134 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
148 135 <span class="i">$NewFPSDFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
149 136 <span class="s">}</span>
|
|
150 137 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
151 138 <span class="i">$NewFPTextFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
152 139 <span class="s">}</span>
|
|
153 140 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
154 141 <span class="i">$NewFPFileIO</span><span class="i">->Close</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
155 142 <span class="s">}</span>
|
|
156 143
|
|
157 144 <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>
|
|
158 145 <span class="s">}</span>
|
|
159 146
|
|
160 147 <span class="c"># Process compound being ignored due to problems in fingerprints geneation...</span>
|
|
161 148 <span class="c">#</span>
|
|
162 <a name="ProcessIgnoredCompound-"></a> 149 <span class="k">sub </span><span class="m">ProcessIgnoredCompound</span> <span class="s">{</span>
|
|
163 150 <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">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
164 151 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdID</span><span class="cm">,</span> <span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
165 152
|
|
166 153 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="i">$Molecule</span><span class="i">->GetDataFieldLabelAndValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
167 154 <span class="i">$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">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
168 155
|
|
169 156 <span class="j">MODE:</span> <span class="s">{</span>
|
|
170 157 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^ContainsNonElementalData$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
171 158 <span class="k">warn</span> <span class="q">"\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains atom data corresponding to non-elemental atom symbol(s)...\n\n"</span><span class="sc">;</span>
|
|
172 159 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
173 160 <span class="s">}</span>
|
|
174 161
|
|
175 162 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^ContainsNoElementalData$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
176 163 <span class="k">warn</span> <span class="q">"\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains no atom data...\n\n"</span><span class="sc">;</span>
|
|
177 164 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
178 165 <span class="s">}</span>
|
|
179 166
|
|
180 167 <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^FingerprintsGenerationFailed$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
181 168 <span class="k">warn</span> <span class="q">"\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn't succeed...\n\n"</span><span class="sc">;</span>
|
|
182 169 <span class="k">next</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
183 170 <span class="s">}</span>
|
|
184 171 <span class="k">warn</span> <span class="q">"\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn't succeed...\n\n"</span><span class="sc">;</span>
|
|
185 172 <span class="s">}</span>
|
|
186 173 <span class="s">}</span>
|
|
187 174
|
|
188 175 <span class="c"># Check and filter compounds....</span>
|
|
189 176 <span class="c">#</span>
|
|
190 <a name="CheckAndFilterCompound-"></a> 177 <span class="k">sub </span><span class="m">CheckAndFilterCompound</span> <span class="s">{</span>
|
|
191 178 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
192 179 <span class="k">my</span><span class="s">(</span><span class="i">$ElementCount</span><span class="cm">,</span> <span class="i">$NonElementCount</span><span class="s">)</span><span class="sc">;</span>
|
|
193 180
|
|
194 181 <span class="s">(</span><span class="i">$ElementCount</span><span class="cm">,</span> <span class="i">$NonElementCount</span><span class="s">)</span> = <span class="i">$Molecule</span><span class="i">->GetNumOfElementsAndNonElements</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
195 182
|
|
196 183 <span class="k">if</span> <span class="s">(</span><span class="i">$NonElementCount</span><span class="s">)</span> <span class="s">{</span>
|
|
197 184 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">'ContainsNonElementalData'</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="s">)</span><span class="sc">;</span>
|
|
198 185 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
199 186 <span class="s">}</span>
|
|
200 187
|
|
201 188 <span class="k">if</span> <span class="s">(</span>!<span class="i">$ElementCount</span><span class="s">)</span> <span class="s">{</span>
|
|
202 189 <span class="i">ProcessIgnoredCompound</span><span class="s">(</span><span class="q">'ContainsNoElementalData'</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>
|
|
203 190 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
204 191 <span class="s">}</span>
|
|
205 192
|
|
206 193 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
207 194 <span class="s">}</span>
|
|
208 195
|
|
209 196 <span class="c"># Write out compounds fingerprints generation summary statistics...</span>
|
|
210 197 <span class="c">#</span>
|
|
211 <a name="WriteFingerprintsGenerationSummaryStatistics-"></a> 198 <span class="k">sub </span><span class="m">WriteFingerprintsGenerationSummaryStatistics</span> <span class="s">{</span>
|
|
212 199 <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>
|
|
213 200 <span class="k">my</span><span class="s">(</span><span class="i">$ProcessedCmpdCount</span><span class="s">)</span><span class="sc">;</span>
|
|
214 201
|
|
215 202 <span class="i">$ProcessedCmpdCount</span> = <span class="i">$CmpdCount</span> - <span class="i">$IgnoredCmpdCount</span><span class="sc">;</span>
|
|
216 203
|
|
217 204 <span class="k">print</span> <span class="q">"\nNumber of compounds: $CmpdCount\n"</span><span class="sc">;</span>
|
|
218 205 <span class="k">print</span> <span class="q">"Number of compounds processed successfully during fingerprints generation: $ProcessedCmpdCount\n"</span><span class="sc">;</span>
|
|
219 206 <span class="k">print</span> <span class="q">"Number of compounds ignored during fingerprints generation: $IgnoredCmpdCount\n"</span><span class="sc">;</span>
|
|
220 207 <span class="s">}</span>
|
|
221 208
|
|
222 209 <span class="c"># Open output files...</span>
|
|
223 210 <span class="c">#</span>
|
|
224 <a name="SetupAndOpenOutputFiles-"></a> 211 <span class="k">sub </span><span class="m">SetupAndOpenOutputFiles</span> <span class="s">{</span>
|
|
225 212 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
226 213 <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>
|
|
227 214
|
|
228 215 <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">$NewFPFileIO</span><span class="s">)</span> = <span class="s">(</span><span class="k">undef</span><span class="s">)</span> x <span class="n">3</span><span class="sc">;</span>
|
|
229 216
|
|
230 217 <span class="c"># Setup common parameters for fingerprints file IO objects...</span>
|
|
231 218 <span class="c">#</span>
|
|
232 219 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
233 220 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^(ExtendedConnectivity|ExtendedConnectivityCount)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
234 221 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">'Mode'</span> <span class="cm">=></span> <span class="q">'Write'</span><span class="cm">,</span> <span class="q">'Overwrite'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="cm">,</span> <span class="q">'FingerprintsStringMode'</span> <span class="cm">=></span> <span class="q">'FingerprintsVectorString'</span><span class="cm">,</span> <span class="q">'VectorStringFormat'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">VectorStringFormat</span>}<span class="s">)</span><span class="sc">;</span>
|
|
235 222 <span class="s">}</span>
|
|
236 223 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^ExtendedConnectivityBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
237 224 <span class="i">%FingerprintsFileIOParams</span> = <span class="s">(</span><span class="q">'Mode'</span> <span class="cm">=></span> <span class="q">'Write'</span><span class="cm">,</span> <span class="q">'Overwrite'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="cm">,</span> <span class="q">'FingerprintsStringMode'</span> <span class="cm">=></span> <span class="q">'FingerprintsBitVectorString'</span><span class="cm">,</span> <span class="q">'BitStringFormat'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">BitStringFormat</span>}<span class="cm">,</span> <span class="q">'BitsOrder'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">BitsOrder</span>}<span class="s">)</span><span class="sc">;</span>
|
|
238 225 <span class="s">}</span>
|
|
239 226
|
|
240 227 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
241 228 <span class="i">$NewFPSDFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
242 229 <span class="k">print</span> <span class="q">"Generating SD file $NewFPSDFile...\n"</span><span class="sc">;</span>
|
|
243 230 <span class="i">$NewFPSDFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsSDFileIO</span><span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$NewFPSDFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="cm">,</span> <span class="q">'FingerprintsFieldLabel'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>}<span class="s">)</span><span class="sc">;</span>
|
|
244 231 <span class="i">$NewFPSDFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
245 232 <span class="s">}</span>
|
|
246 233
|
|
247 234 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">FPOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
248 235 <span class="i">$NewFPFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
249 236 <span class="k">print</span> <span class="q">"Generating FP file $NewFPFile...\n"</span><span class="sc">;</span>
|
|
250 237 <span class="i">$NewFPFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsFPFileIO</span><span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$NewFPFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="s">)</span><span class="sc">;</span>
|
|
251 238 <span class="i">$NewFPFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
252 239 <span class="s">}</span>
|
|
253 240
|
|
254 241 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
255 242 <span class="k">my</span><span class="s">(</span><span class="i">$ColLabelsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
256 243
|
|
257 244 <span class="i">$NewFPTextFile</span> = <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
258 245 <span class="i">$ColLabelsRef</span> = <span class="i">SetupFPTextFileCoulmnLabels</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
259 246
|
|
260 247 <span class="k">print</span> <span class="q">"Generating text file $NewFPTextFile...\n"</span><span class="sc">;</span>
|
|
261 248 <span class="i">$NewFPTextFileIO</span> = <span class="w">new</span> <span class="i">FileIO::FingerprintsTextFileIO</span><span class="s">(</span><span class="q">'Name'</span> <span class="cm">=></span> <span class="i">$NewFPTextFile</span><span class="cm">,</span> <span class="i">%FingerprintsFileIOParams</span><span class="cm">,</span> <span class="q">'DataColLabels'</span> <span class="cm">=></span> <span class="i">$ColLabelsRef</span><span class="cm">,</span> <span class="q">'OutDelim'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="q">'OutQuote'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
262 249 <span class="i">$NewFPTextFileIO</span><span class="i">->Open</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
263 250 <span class="s">}</span>
|
|
264 251
|
|
265 252 <span class="k">return</span> <span class="s">(</span><span class="i">$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>
|
|
266 253 <span class="s">}</span>
|
|
267 254
|
|
268 255 <span class="c"># Write fingerpritns and other data to appropriate output files...</span>
|
|
269 256 <span class="c">#</span>
|
|
270 <a name="WriteDataToOutputFiles-"></a> 257 <span class="k">sub </span><span class="m">WriteDataToOutputFiles</span> <span class="s">{</span>
|
|
271 258 <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">$Molecule</span><span class="cm">,</span> <span class="i">$ExtendedConnectivityFingerprints</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">$NewFPFileIO</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
272 259 <span class="k">my</span><span class="s">(</span><span class="i">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
273 260
|
|
274 261 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="k">undef</span><span class="sc">;</span>
|
|
275 262 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span> || <span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
276 263 <span class="i">$DataFieldLabelAndValuesRef</span> = <span class="i">$Molecule</span><span class="i">->GetDataFieldLabelAndValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
277 264 <span class="s">}</span>
|
|
278 265
|
|
279 266 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPSDFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
280 267 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdString</span><span class="s">)</span><span class="sc">;</span>
|
|
281 268
|
|
282 269 <span class="i">$CmpdString</span> = <span class="i">$Molecule</span><span class="i">->GetInputMoleculeString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
283 270 <span class="i">$NewFPSDFileIO</span><span class="i">->WriteFingerprints</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="s">)</span><span class="sc">;</span>
|
|
284 271 <span class="s">}</span>
|
|
285 272
|
|
286 273 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPTextFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
287 274 <span class="k">my</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
288 275
|
|
289 276 <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">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
290 277 <span class="i">$NewFPTextFileIO</span><span class="i">->WriteFingerprints</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="cm">,</span> <span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
291 278 <span class="s">}</span>
|
|
292 279
|
|
293 280 <span class="k">if</span> <span class="s">(</span><span class="i">$NewFPFileIO</span><span class="s">)</span> <span class="s">{</span>
|
|
294 281 <span class="k">my</span><span class="s">(</span><span class="i">$CompoundID</span><span class="s">)</span><span class="sc">;</span>
|
|
295 282
|
|
296 283 <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>
|
|
297 284 <span class="i">$NewFPFileIO</span><span class="i">->WriteFingerprints</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="cm">,</span> <span class="i">$CompoundID</span><span class="s">)</span><span class="sc">;</span>
|
|
298 285 <span class="s">}</span>
|
|
299 286 <span class="s">}</span>
|
|
300 287
|
|
301 288 <span class="c"># Generate approriate column labels for FPText output file...</span>
|
|
302 289 <span class="c">#</span>
|
|
303 <a name="SetupFPTextFileCoulmnLabels-"></a> 290 <span class="k">sub </span><span class="m">SetupFPTextFileCoulmnLabels</span> <span class="s">{</span>
|
|
304 291 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
305 292 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="s">)</span><span class="sc">;</span>
|
|
306 293
|
|
307 294 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
308 295 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
309 296 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
310 297 <span class="s">}</span>
|
|
311 298 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
312 299 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
313 300 <span class="s">}</span>
|
|
314 301 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
315 302 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="sc">;</span>
|
|
316 303 <span class="s">}</span>
|
|
317 304 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
318 305 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDLabel</span>}<span class="sc">;</span>
|
|
319 306 <span class="s">}</span>
|
|
320 307 <span class="c"># Add fingerprints label...</span>
|
|
321 308 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>}<span class="sc">;</span>
|
|
322 309
|
|
323 310 <span class="k">return</span> \<span class="i">@ColLabels</span><span class="sc">;</span>
|
|
324 311 <span class="s">}</span>
|
|
325 312
|
|
326 313 <span class="c"># Generate column values FPText output file..</span>
|
|
327 314 <span class="c">#</span>
|
|
328 <a name="SetupFPTextFileCoulmnValues-"></a> 315 <span class="k">sub </span><span class="m">SetupFPTextFileCoulmnValues</span> <span class="s">{</span>
|
|
329 316 <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>
|
|
330 317 <span class="k">my</span><span class="s">(</span><span class="i">@ColValues</span><span class="s">)</span><span class="sc">;</span>
|
|
331 318
|
|
332 319 <span class="i">@ColValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
333 320 <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>
|
|
334 321 <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">$DataFieldLabelAndValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
335 322 <span class="s">}</span>
|
|
336 323 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
337 324 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} <span class="co">:</span> <span class="q">''</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
338 325 <span class="s">}</span>
|
|
339 326 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Common$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
340 327 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} <span class="co">:</span> <span class="q">''</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
341 328 <span class="s">}</span>
|
|
342 329 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
343 330 <span class="i">@ColValues</span> = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$_</span>} <span class="co">:</span> <span class="q">''</span><span class="s">}</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="sc">;</span>
|
|
344 331 <span class="s">}</span>
|
|
345 332
|
|
346 333 <span class="k">return</span> \<span class="i">@ColValues</span><span class="sc">;</span>
|
|
347 334 <span class="s">}</span>
|
|
348 335
|
|
349 336 <span class="c"># Generate compound ID for FP and FPText output files..</span>
|
|
350 337 <span class="c">#</span>
|
|
351 <a name="SetupCmpdIDForOutputFiles-"></a> 338 <span class="k">sub </span><span class="m">SetupCmpdIDForOutputFiles</span> <span class="s">{</span>
|
|
352 339 <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">$DataFieldLabelAndValuesRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
353 340 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdID</span><span class="s">)</span><span class="sc">;</span>
|
|
354 341
|
|
355 342 <span class="i">$CmpdID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
356 343 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^MolNameOrLabelPrefix$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
357 344 <span class="k">my</span><span class="s">(</span><span class="i">$MolName</span><span class="s">)</span><span class="sc">;</span>
|
|
358 345 <span class="i">$MolName</span> = <span class="i">$Molecule</span><span class="i">->GetName</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
359 346 <span class="i">$CmpdID</span> = <span class="i">$MolName</span> ? <span class="i">$MolName</span> <span class="co">:</span> <span class="q">"$OptionsInfo{CompoundID}${CmpdCount}"</span><span class="sc">;</span>
|
|
360 347 <span class="s">}</span>
|
|
361 348 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^LabelPrefix$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
362 349 <span class="i">$CmpdID</span> = <span class="q">"$OptionsInfo{CompoundID}${CmpdCount}"</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">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
365 352 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedDataField</span><span class="s">)</span><span class="sc">;</span>
|
|
366 353 <span class="i">$SpecifiedDataField</span> = <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>}<span class="sc">;</span>
|
|
367 354 <span class="i">$CmpdID</span> = <span class="k">exists</span> <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$SpecifiedDataField</span>} ? <span class="i">$DataFieldLabelAndValuesRef</span>->{<span class="i">$SpecifiedDataField</span>} <span class="co">:</span> <span class="q">''</span><span class="sc">;</span>
|
|
368 355 <span class="s">}</span>
|
|
369 356 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} =~ <span class="q">/^MolName$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
370 357 <span class="i">$CmpdID</span> = <span class="i">$Molecule</span><span class="i">->GetName</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
371 358 <span class="s">}</span>
|
|
372 359 <span class="k">return</span> <span class="i">$CmpdID</span><span class="sc">;</span>
|
|
373 360 <span class="s">}</span>
|
|
374 361
|
|
375 362 <span class="c"># Generate fingerprints for molecule...</span>
|
|
376 363 <span class="c">#</span>
|
|
377 <a name="GenerateMoleculeFingerprints-"></a> 364 <span class="k">sub </span><span class="m">GenerateMoleculeFingerprints</span> <span class="s">{</span>
|
|
378 365 <span class="k">my</span><span class="s">(</span><span class="i">$Molecule</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
379 366 <span class="k">my</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="s">)</span><span class="sc">;</span>
|
|
380 367
|
|
381 368 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">KeepLargestComponent</span>}<span class="s">)</span> <span class="s">{</span>
|
|
382 369 <span class="i">$Molecule</span><span class="i">->KeepLargestComponent</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
383 370 <span class="s">}</span>
|
|
384 371 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Molecule</span><span class="i">->DetectRings</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
385 372 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
386 373 <span class="s">}</span>
|
|
387 374 <span class="i">$Molecule</span><span class="i">->SetAromaticityModel</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AromaticityModel</span>}<span class="s">)</span><span class="sc">;</span>
|
|
388 375 <span class="i">$Molecule</span><span class="i">->DetectAromaticity</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
389 376
|
|
390 377 <span class="i">$ExtendedConnectivityFingerprints</span> = <span class="k">undef</span><span class="sc">;</span>
|
|
391 378 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^(ExtendedConnectivity|ExtendedConnectivityCount)$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
392 379 <span class="i">$ExtendedConnectivityFingerprints</span> = <span class="w">new</span> <span class="i">Fingerprints::ExtendedConnectivityFingerprints</span><span class="s">(</span><span class="q">'Type'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>}<span class="cm">,</span> <span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="q">'NeighborhoodRadius'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">NeighborhoodRadius</span>}<span class="cm">,</span> <span class="q">'AtomIdentifierType'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>}<span class="s">)</span><span class="sc">;</span>
|
|
393 380 <span class="s">}</span>
|
|
394 381 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^ExtendedConnectivityBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
395 382 <span class="i">$ExtendedConnectivityFingerprints</span> = <span class="w">new</span> <span class="i">Fingerprints::ExtendedConnectivityFingerprints</span><span class="s">(</span><span class="q">'Type'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>}<span class="cm">,</span> <span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="q">'NeighborhoodRadius'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">NeighborhoodRadius</span>}<span class="cm">,</span> <span class="q">'AtomIdentifierType'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>}<span class="cm">,</span> <span class="q">'Size'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">Size</span>}<span class="cm">,</span> <span class="q">'UsePerlCoreRandom'</span> <span class="cm">=></span> <span class="i">$OptionsInfo</span>{<span class="w">UsePerlCoreRandom</span>}<span class="s">)</span><span class="sc">;</span>
|
|
396 383 <span class="s">}</span>
|
|
397 384 <span class="k">else</span> <span class="s">{</span>
|
|
398 385 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: ExtendedConnectivity, ExtendedConnectivityCount or ExtendedConnectivityBits\n"</span><span class="sc">;</span>
|
|
399 386 <span class="s">}</span>
|
|
400 387 <span class="i">SetAtomIdentifierTypeValuesToUse</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="s">)</span><span class="sc">;</span>
|
|
401 388
|
|
402 389 <span class="c"># Generate fingerprints...</span>
|
|
403 390 <span class="i">$ExtendedConnectivityFingerprints</span><span class="i">->GenerateFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
404 391
|
|
405 392 <span class="c"># Make sure fingerprints generation is successful...</span>
|
|
406 393 <span class="k">if</span> <span class="s">(</span>!<span class="i">$ExtendedConnectivityFingerprints</span><span class="i">->IsFingerprintsGenerationSuccessful</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
407 394 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
408 395 <span class="s">}</span>
|
|
409 396
|
|
410 397 <span class="k">return</span> <span class="i">$ExtendedConnectivityFingerprints</span><span class="sc">;</span>
|
|
411 398 <span class="s">}</span>
|
|
412 399
|
|
413 400 <span class="c"># Set atom identifier type to use for generating fingerprints...</span>
|
|
414 401 <span class="c">#</span>
|
|
415 <a name="SetAtomIdentifierTypeValuesToUse-"></a> 402 <span class="k">sub </span><span class="m">SetAtomIdentifierTypeValuesToUse</span> <span class="s">{</span>
|
|
416 403 <span class="k">my</span><span class="s">(</span><span class="i">$ExtendedConnectivityFingerprints</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
417 404
|
|
418 405 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^AtomicInvariantsAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
419 406 <span class="i">$ExtendedConnectivityFingerprints</span><span class="i">->SetAtomicInvariantsToUse</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">AtomicInvariantsToUse</span>}}<span class="s">)</span><span class="sc">;</span>
|
|
420 407 <span class="s">}</span>
|
|
421 408 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
422 409 <span class="i">$ExtendedConnectivityFingerprints</span><span class="i">->SetFunctionalClassesToUse</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">FunctionalClassesToUse</span>}}<span class="s">)</span><span class="sc">;</span>
|
|
423 410 <span class="s">}</span>
|
|
424 411 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^(DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
425 412 <span class="c"># Nothing to do for now...</span>
|
|
426 413 <span class="s">}</span>
|
|
427 414 <span class="k">else</span> <span class="s">{</span>
|
|
428 415 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n"</span><span class="sc">;</span>
|
|
429 416 <span class="s">}</span>
|
|
430 417 <span class="s">}</span>
|
|
431 418
|
|
432 419 <span class="c"># Retrieve information about SD files...</span>
|
|
433 420 <span class="c">#</span>
|
|
434 <a name="RetrieveSDFilesInfo-"></a> 421 <span class="k">sub </span><span class="m">RetrieveSDFilesInfo</span> <span class="s">{</span>
|
|
435 422 <span class="k">my</span><span class="s">(</span><span class="i">$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">$CommonDataFieldsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
436 423
|
|
437 424 <span class="i">%SDFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
438 425 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
439 426 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
440 427 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
441 428 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
442 429 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
443 430 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
444 431 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
445 432
|
|
446 433 <span class="i">$CheckDataField</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</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><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>
|
|
447 434 <span class="i">$CollectDataFields</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>} && <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>
|
|
448 435
|
|
449 436 <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>
|
|
450 437 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
451 438
|
|
452 439 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
453 440 <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
454 441 <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
455 442 <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
456 443 <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
457 444
|
|
458 445 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
459 446 <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>
|
|
460 447 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It doesn't exist\n"</span><span class="sc">;</span>
|
|
461 448 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
462 449 <span class="s">}</span>
|
|
463 450 <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="q">"sd sdf"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
464 451 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It's not a SD file\n"</span><span class="sc">;</span>
|
|
465 452 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
466 453 <span class="s">}</span>
|
|
467 454
|
|
468 455 <span class="k">if</span> <span class="s">(</span><span class="i">$CheckDataField</span><span class="s">)</span> <span class="s">{</span>
|
|
469 456 <span class="c"># Make sure data field exists in SD file..</span>
|
|
470 457 <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>
|
|
471 458
|
|
472 459 <span class="i">@CmpdLines</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
473 460 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $SDFile: $! \n"</span><span class="sc">;</span>
|
|
474 461 <span class="i">$CmpdString</span> = <span class="i">ReadCmpdString</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
475 462 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
|
|
476 463 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span>
|
|
477 464 <span class="i">%DataFieldValues</span> = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
478 465 <span class="i">$SpecifiedDataField</span> = <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>}<span class="sc">;</span>
|
|
479 466 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$DataFieldValues</span>{<span class="i">$SpecifiedDataField</span>}<span class="s">)</span> <span class="s">{</span>
|
|
480 467 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Data field value, $SpecifiedDataField, using \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\" doesn't exist\n"</span><span class="sc">;</span>
|
|
481 468 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
482 469 <span class="s">}</span>
|
|
483 470 <span class="s">}</span>
|
|
484 471
|
|
485 472 <span class="i">$AllDataFieldsRef</span> = <span class="q">''</span><span class="sc">;</span>
|
|
486 473 <span class="i">$CommonDataFieldsRef</span> = <span class="q">''</span><span class="sc">;</span>
|
|
487 474 <span class="k">if</span> <span class="s">(</span><span class="i">$CollectDataFields</span><span class="s">)</span> <span class="s">{</span>
|
|
488 475 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdCount</span><span class="s">)</span><span class="sc">;</span>
|
|
489 476 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $SDFile: $! \n"</span><span class="sc">;</span>
|
|
490 477 <span class="s">(</span><span class="i">$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>
|
|
491 478 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
|
|
492 479 <span class="s">}</span>
|
|
493 480
|
|
494 481 <span class="c"># Setup output file names...</span>
|
|
495 482 <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">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
496 483 <span class="s">(</span><span class="i">$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">$SDFile</span><span class="s">)</span><span class="sc">;</span>
|
|
497 484
|
|
498 485 <span class="i">$TextOutFileExt</span> = <span class="q">"csv"</span><span class="sc">;</span>
|
|
499 486 <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>
|
|
500 487 <span class="i">$TextOutFileExt</span> = <span class="q">"tsv"</span><span class="sc">;</span>
|
|
501 488 <span class="s">}</span>
|
|
502 489 <span class="i">$SDOutFileExt</span> = <span class="i">$FileExt</span><span class="sc">;</span>
|
|
503 490 <span class="i">$FPOutFileExt</span> = <span class="q">"fpf"</span><span class="sc">;</span>
|
|
504 491
|
|
505 492 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} && <span class="s">(</span><span class="i">@SDFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
506 493 <span class="k">my</span> <span class="s">(</span><span class="i">$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">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span>
|
|
507 494 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
508 495 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
509 496 <span class="s">}</span>
|
|
510 497 <span class="k">else</span> <span class="s">{</span>
|
|
511 498 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
|
|
512 499 <span class="s">}</span>
|
|
513 500 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
514 501 <span class="s">}</span>
|
|
515 502 <span class="k">else</span> <span class="s">{</span>
|
|
516 503 <span class="i">$OutFileRoot</span> = <span class="q">"${FileName}ExtendedConnectivityFP"</span><span class="sc">;</span>
|
|
517 504 <span class="s">}</span>
|
|
518 505
|
|
519 506 <span class="i">$NewSDFileName</span> = <span class="q">"${OutFileRoot}.${SDOutFileExt}"</span><span class="sc">;</span>
|
|
520 507 <span class="i">$NewFPFileName</span> = <span class="q">"${OutFileRoot}.${FPOutFileExt}"</span><span class="sc">;</span>
|
|
521 508 <span class="i">$NewTextFileName</span> = <span class="q">"${OutFileRoot}.${TextOutFileExt}"</span><span class="sc">;</span>
|
|
522 509
|
|
523 510 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
524 511 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFile</span> =~ <span class="q">/$NewSDFileName/i</span><span class="s">)</span> <span class="s">{</span>
|
|
525 512 <span class="k">warn</span> <span class="q">"Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n"</span><span class="sc">;</span>
|
|
526 513 <span class="k">print</span> <span class="q">"Specify a different name using \"-r --root\" option or use default name.\n"</span><span class="sc">;</span>
|
|
527 514 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
528 515 <span class="s">}</span>
|
|
529 516 <span class="s">}</span>
|
|
530 517
|
|
531 518 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span>
|
|
532 519 <span class="c"># Check SD and text outout files...</span>
|
|
533 520 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SDOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
534 521 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewSDFileName</span><span class="s">)</span> <span class="s">{</span>
|
|
535 522 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: The file $NewSDFileName already exists\n"</span><span class="sc">;</span>
|
|
536 523 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
537 524 <span class="s">}</span>
|
|
538 525 <span class="s">}</span>
|
|
539 526 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">FPOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
540 527 <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>
|
|
541 528 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: The file $NewFPFileName already exists\n"</span><span class="sc">;</span>
|
|
542 529 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
543 530 <span class="s">}</span>
|
|
544 531 <span class="s">}</span>
|
|
545 532 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">TextOutput</span>}<span class="s">)</span> <span class="s">{</span>
|
|
546 533 <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>
|
|
547 534 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: The file $NewTextFileName already exists\n"</span><span class="sc">;</span>
|
|
548 535 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
549 536 <span class="s">}</span>
|
|
550 537 <span class="s">}</span>
|
|
551 538 <span class="s">}</span>
|
|
552 539
|
|
553 540 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
554 541
|
|
555 542 <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span>
|
|
556 543 <span class="i">$SDFilesInfo</span>{<span class="w">SDOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewSDFileName</span><span class="sc">;</span>
|
|
557 544 <span class="i">$SDFilesInfo</span>{<span class="w">FPOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewFPFileName</span><span class="sc">;</span>
|
|
558 545 <span class="i">$SDFilesInfo</span>{<span class="w">TextOutFileNames</span>}[<span class="i">$Index</span>] = <span class="i">$NewTextFileName</span><span class="sc">;</span>
|
|
559 546
|
|
560 547 <span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldsRef</span>}[<span class="i">$Index</span>] = <span class="i">$AllDataFieldsRef</span><span class="sc">;</span>
|
|
561 548 <span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldsRef</span>}[<span class="i">$Index</span>] = <span class="i">$CommonDataFieldsRef</span><span class="sc">;</span>
|
|
562 549 <span class="s">}</span>
|
|
563 550 <span class="s">}</span>
|
|
564 551
|
|
565 552 <span class="c"># Process option values...</span>
|
|
566 <a name="ProcessOptions-"></a> 553 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
567 554 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
568 555
|
|
569 556 <span class="i">ProcessAtomIdentifierTypeOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
570 557
|
|
571 558 <span class="i">$OptionsInfo</span>{<span class="w">AromaticityModel</span>} = <span class="i">$Options</span>{<span class="w">aromaticitymodel</span>}<span class="sc">;</span>
|
|
572 559
|
|
573 560 <span class="i">$OptionsInfo</span>{<span class="w">BitsOrder</span>} = <span class="i">$Options</span>{<span class="w">bitsorder</span>}<span class="sc">;</span>
|
|
574 561 <span class="i">$OptionsInfo</span>{<span class="w">BitStringFormat</span>} = <span class="i">$Options</span>{<span class="w">bitstringformat</span>}<span class="sc">;</span>
|
|
575 562
|
|
576 563 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDMode</span>} = <span class="i">$Options</span>{<span class="w">compoundidmode</span>}<span class="sc">;</span>
|
|
577 564 <span class="i">$OptionsInfo</span>{<span class="w">CompoundIDLabel</span>} = <span class="i">$Options</span>{<span class="w">compoundidlabel</span>}<span class="sc">;</span>
|
|
578 565 <span class="i">$OptionsInfo</span>{<span class="w">DataFieldsMode</span>} = <span class="i">$Options</span>{<span class="w">datafieldsmode</span>}<span class="sc">;</span>
|
|
579 566
|
|
580 567 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedDataFields</span><span class="s">)</span><span class="sc">;</span>
|
|
581 568 <span class="i">@SpecifiedDataFields</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
582 569
|
|
583 570 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
584 571 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
585 572
|
|
586 573 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} =~ <span class="q">/^CompoundID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
587 574 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
588 575 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">compoundid</span>}<span class="s">)</span> <span class="s">{</span>
|
|
589 576 <span class="k">die</span> <span class="q">"Error: You must specify a value for \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\". \n"</span><span class="sc">;</span>
|
|
590 577 <span class="s">}</span>
|
|
591 578 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="i">$Options</span>{<span class="w">compoundid</span>}<span class="sc">;</span>
|
|
592 579 <span class="s">}</span>
|
|
593 580 <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>
|
|
594 581 <span class="i">$OptionsInfo</span>{<span class="w">CompoundID</span>} = <span class="i">$Options</span>{<span class="w">compoundid</span>} ? <span class="i">$Options</span>{<span class="w">compoundid</span>} <span class="co">:</span> <span class="q">'Cmpd'</span><span class="sc">;</span>
|
|
595 582 <span class="s">}</span>
|
|
596 583 <span class="s">}</span>
|
|
597 584 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} =~ <span class="q">/^Specify$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
598 585 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
599 586 <span class="k">die</span> <span class="q">"Error: You must specify a value for \"--DataFields\" option in \"Specify\" \"-d, --DataFieldsMode\". \n"</span><span class="sc">;</span>
|
|
600 587 <span class="s">}</span>
|
|
601 588 <span class="i">@SpecifiedDataFields</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="sc">;</span>
|
|
602 589 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFields</span>}}<span class="cm">,</span> <span class="i">@SpecifiedDataFields</span><span class="sc">;</span>
|
|
603 590 <span class="s">}</span>
|
|
604 591
|
|
605 592 <span class="i">$OptionsInfo</span>{<span class="w">FingerprintsLabel</span>} = <span class="i">$Options</span>{<span class="w">fingerprintslabel</span>} ? <span class="i">$Options</span>{<span class="w">fingerprintslabel</span>} <span class="co">:</span> <span class="q">'ExtendedConnectivityFingerprints'</span><span class="sc">;</span>
|
|
606 593
|
|
607 594 <span class="i">$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>
|
|
608 595
|
|
609 596 <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$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
610 597
|
|
611 598 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
612 599
|
|
613 600 <span class="i">$OptionsInfo</span>{<span class="w">NeighborhoodRadius</span>} = <span class="i">$Options</span>{<span class="w">neighborhoodradius</span>}<span class="sc">;</span>
|
|
614 601
|
|
615 602 <span class="i">$OptionsInfo</span>{<span class="w">UsePerlCoreRandom</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">useperlcorerandom</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
616 603
|
|
617 604 <span class="i">$OptionsInfo</span>{<span class="w">Output</span>} = <span class="i">$Options</span>{<span class="w">output</span>}<span class="sc">;</span>
|
|
618 605 <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)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
619 606 <span class="i">$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>
|
|
620 607 <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)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
621 608
|
|
622 609 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="i">$Options</span>{<span class="w">outdelim</span>}<span class="sc">;</span>
|
|
623 610 <span class="i">$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>
|
|
624 611
|
|
625 612 <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>
|
|
626 613 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
627 614
|
|
628 615 <span class="k">my</span><span class="s">(</span><span class="i">$Size</span><span class="cm">,</span> <span class="i">$MinSize</span><span class="cm">,</span> <span class="i">$MaxSize</span><span class="s">)</span><span class="sc">;</span>
|
|
629 616 <span class="i">$MinSize</span> = <span class="n">32</span><span class="sc">;</span>
|
|
630 617 <span class="i">$MaxSize</span> = <span class="n">2</span>**<span class="n">32</span><span class="sc">;</span>
|
|
631 618 <span class="i">$Size</span> = <span class="i">$Options</span>{<span class="w">size</span>}<span class="sc">;</span>
|
|
632 619 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Size</span><span class="s">)</span> && <span class="i">$Size</span> >= <span class="i">$MinSize</span> && <span class="i">$Size</span> <= <span class="i">$MaxSize</span> && <span class="i">IsNumberPowerOfNumber</span><span class="s">(</span><span class="i">$Size</span><span class="cm">,</span> <span class="n">2</span><span class="s">)</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
633 620 <span class="k">die</span> <span class="q">"Error: Invalid size value, $Size, for \"-s, --size\" option. Allowed values: power of 2, >= minimum size of $MinSize, and <= maximum size of $MaxSize.\n"</span><span class="sc">;</span>
|
|
634 621 <span class="s">}</span>
|
|
635 622 <span class="i">$OptionsInfo</span>{<span class="w">Size</span>} = <span class="i">$Size</span><span class="sc">;</span>
|
|
636 623
|
|
637 624 <span class="c"># Setup default vector string format...</span>
|
|
638 625 <span class="c">#</span>
|
|
639 626 <span class="k">my</span><span class="s">(</span><span class="i">$VectorStringFormat</span><span class="s">)</span><span class="sc">;</span>
|
|
640 627 <span class="i">$VectorStringFormat</span> = <span class="q">''</span><span class="sc">;</span>
|
|
641 628 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">vectorstringformat</span>}<span class="s">)</span> <span class="s">{</span>
|
|
642 629 <span class="i">$VectorStringFormat</span> = <span class="i">$Options</span>{<span class="w">vectorstringformat</span>}<span class="sc">;</span>
|
|
643 630 <span class="s">}</span>
|
|
644 631 <span class="k">else</span> <span class="s">{</span>
|
|
645 632 <span class="i">$VectorStringFormat</span> = <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^ExtendedConnectivity$/</span><span class="s">)</span> ? <span class="q">"ValuesString"</span> <span class="co">:</span> <span class="q">"IDsAndValuesString"</span><span class="sc">;</span>
|
|
646 633 <span class="s">}</span>
|
|
647 634 <span class="i">$OptionsInfo</span>{<span class="w">VectorStringFormat</span>} = <span class="i">$VectorStringFormat</span><span class="sc">;</span>
|
|
648 635 <span class="s">}</span>
|
|
649 636
|
|
650 637 <span class="c"># Process atom identifier type and related options...</span>
|
|
651 638 <span class="c">#</span>
|
|
652 <a name="ProcessAtomIdentifierTypeOptions-"></a> 639 <span class="k">sub </span><span class="m">ProcessAtomIdentifierTypeOptions</span> <span class="s">{</span>
|
|
653 640
|
|
654 641 <span class="i">$OptionsInfo</span>{<span class="w">AtomIdentifierType</span>} = <span class="i">$Options</span>{<span class="w">atomidentifiertype</span>}<span class="sc">;</span>
|
|
655 642
|
|
656 643 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">atomidentifiertype</span>} =~ <span class="q">/^AtomicInvariantsAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
657 644 <span class="i">ProcessAtomicInvariantsToUseOption</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
658 645 <span class="s">}</span>
|
|
659 646 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">atomidentifiertype</span>} =~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
660 647 <span class="i">ProcessFunctionalClassesToUse</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
661 648 <span class="s">}</span>
|
|
662 649 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">atomidentifiertype</span>} =~ <span class="q">/^(DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
663 650 <span class="c"># Nothing to do for now...</span>
|
|
664 651 <span class="s">}</span>
|
|
665 652 <span class="k">else</span> <span class="s">{</span>
|
|
666 653 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n"</span><span class="sc">;</span>
|
|
667 654 <span class="s">}</span>
|
|
668 655 <span class="s">}</span>
|
|
669 656
|
|
670 657 <span class="c"># Process specified atomic invariants to use...</span>
|
|
671 658 <span class="c">#</span>
|
|
672 <a name="ProcessAtomicInvariantsToUseOption-"></a> 659 <span class="k">sub </span><span class="m">ProcessAtomicInvariantsToUseOption</span> <span class="s">{</span>
|
|
673 660 <span class="k">my</span><span class="s">(</span><span class="i">$AtomicInvariant</span><span class="cm">,</span> <span class="i">$AtomSymbolSpecified</span><span class="cm">,</span> <span class="i">@AtomicInvariantsWords</span><span class="s">)</span><span class="sc">;</span>
|
|
674 661
|
|
675 662 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">AtomicInvariantsToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
676 663 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">atomicinvariantstouse</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
677 664 <span class="k">die</span> <span class="q">"Error: Atomic invariants value specified using \"--AtomicInvariantsToUse\" option is empty\n"</span><span class="sc">;</span>
|
|
678 665 <span class="s">}</span>
|
|
679 666 <span class="i">$AtomSymbolSpecified</span> = <span class="n">0</span><span class="sc">;</span>
|
|
680 667 <span class="i">@AtomicInvariantsWords</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">atomicinvariantstouse</span>}<span class="sc">;</span>
|
|
681 668 <span class="k">for</span> <span class="i">$AtomicInvariant</span> <span class="s">(</span><span class="i">@AtomicInvariantsWords</span><span class="s">)</span> <span class="s">{</span>
|
|
682 669 <span class="k">if</span> <span class="s">(</span>!<span class="i">AtomTypes::AtomicInvariantsAtomTypes::IsAtomicInvariantAvailable</span><span class="s">(</span><span class="i">$AtomicInvariant</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
683 670 <span class="k">die</span> <span class="q">"Error: Atomic invariant specified, $AtomicInvariant, using \"--AtomicInvariantsToUse\" option is not valid...\n "</span><span class="sc">;</span>
|
|
684 671 <span class="s">}</span>
|
|
685 672 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomicInvariant</span> =~ <span class="q">/^(AS|AtomSymbol)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
686 673 <span class="i">$AtomSymbolSpecified</span> = <span class="n">1</span><span class="sc">;</span>
|
|
687 674 <span class="s">}</span>
|
|
688 675 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">AtomicInvariantsToUse</span>}}<span class="cm">,</span> <span class="i">$AtomicInvariant</span><span class="sc">;</span>
|
|
689 676 <span class="s">}</span>
|
|
690 677 <span class="k">if</span> <span class="s">(</span>!<span class="i">$AtomSymbolSpecified</span><span class="s">)</span> <span class="s">{</span>
|
|
691 678 <span class="k">die</span> <span class="q">"Error: Atomic invariant, AS or AtomSymbol, must be specified as using \"--AtomicInvariantsToUse\" option...\n "</span><span class="sc">;</span>
|
|
692 679 <span class="s">}</span>
|
|
693 680 <span class="s">}</span>
|
|
694 681
|
|
695 682 <span class="c"># Process specified functional classes invariants to use...</span>
|
|
696 683 <span class="c">#</span>
|
|
697 <a name="ProcessFunctionalClassesToUse-"></a> 684 <span class="k">sub </span><span class="m">ProcessFunctionalClassesToUse</span> <span class="s">{</span>
|
|
698 685 <span class="k">my</span><span class="s">(</span><span class="i">$FunctionalClass</span><span class="cm">,</span> <span class="i">@FunctionalClassesToUseWords</span><span class="s">)</span><span class="sc">;</span>
|
|
699 686
|
|
700 687 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">FunctionalClassesToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
701 688 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">functionalclassestouse</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
702 689 <span class="k">die</span> <span class="q">"Error: Functional classes value specified using \"--FunctionalClassesToUse\" option is empty\n"</span><span class="sc">;</span>
|
|
703 690 <span class="s">}</span>
|
|
704 691 <span class="i">@FunctionalClassesToUseWords</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">functionalclassestouse</span>}<span class="sc">;</span>
|
|
705 692 <span class="k">for</span> <span class="i">$FunctionalClass</span> <span class="s">(</span><span class="i">@FunctionalClassesToUseWords</span><span class="s">)</span> <span class="s">{</span>
|
|
706 693 <span class="k">if</span> <span class="s">(</span>!<span class="i">AtomTypes::FunctionalClassAtomTypes::IsFunctionalClassAvailable</span><span class="s">(</span><span class="i">$FunctionalClass</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
707 694 <span class="k">die</span> <span class="q">"Error: Functional class specified, $FunctionalClass, using \"--FunctionalClassesToUse\" option is not valid...\n "</span><span class="sc">;</span>
|
|
708 695 <span class="s">}</span>
|
|
709 696 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">FunctionalClassesToUse</span>}}<span class="cm">,</span> <span class="i">$FunctionalClass</span><span class="sc">;</span>
|
|
710 697 <span class="s">}</span>
|
|
711 698 <span class="s">}</span>
|
|
712 699
|
|
713 700 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
714 <a name="SetupScriptUsage-"></a> 701 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
715 702
|
|
716 703 <span class="c"># Retrieve all the options...</span>
|
|
717 704 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
718 705
|
|
719 706 <span class="i">$Options</span>{<span class="w">aromaticitymodel</span>} = <span class="q">'MayaChemToolsAromaticityModel'</span><span class="sc">;</span>
|
|
720 707
|
|
721 708 <span class="i">$Options</span>{<span class="w">atomidentifiertype</span>} = <span class="q">'AtomicInvariantsAtomTypes'</span><span class="sc">;</span>
|
|
722 709 <span class="i">$Options</span>{<span class="w">atomicinvariantstouse</span>} = <span class="q">'AS,X,BO,H,FC,MN'</span><span class="sc">;</span>
|
|
723 710 <span class="i">$Options</span>{<span class="w">functionalclassestouse</span>} = <span class="q">'HBD,HBA,PI,NI,Ar,Hal'</span><span class="sc">;</span>
|
|
724 711
|
|
725 712 <span class="i">$Options</span>{<span class="w">bitsorder</span>} = <span class="q">'Ascending'</span><span class="sc">;</span>
|
|
726 713 <span class="i">$Options</span>{<span class="w">bitstringformat</span>} = <span class="q">'HexadecimalString'</span><span class="sc">;</span>
|
|
727 714
|
|
728 715 <span class="i">$Options</span>{<span class="w">compoundidmode</span>} = <span class="q">'LabelPrefix'</span><span class="sc">;</span>
|
|
729 716 <span class="i">$Options</span>{<span class="w">compoundidlabel</span>} = <span class="q">'CompoundID'</span><span class="sc">;</span>
|
|
730 717 <span class="i">$Options</span>{<span class="w">datafieldsmode</span>} = <span class="q">'CompoundID'</span><span class="sc">;</span>
|
|
731 718
|
|
732 719 <span class="i">$Options</span>{<span class="w">filter</span>} = <span class="q">'Yes'</span><span class="sc">;</span>
|
|
733 720
|
|
734 721 <span class="i">$Options</span>{<span class="w">keeplargestcomponent</span>} = <span class="q">'Yes'</span><span class="sc">;</span>
|
|
735 722
|
|
736 723 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">'ExtendedConnectivity'</span><span class="sc">;</span>
|
|
737 724
|
|
738 725 <span class="i">$Options</span>{<span class="w">neighborhoodradius</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
739 726
|
|
740 727 <span class="i">$Options</span>{<span class="w">useperlcorerandom</span>} = <span class="q">'yes'</span><span class="sc">;</span>
|
|
741 728
|
|
742 729 <span class="i">$Options</span>{<span class="w">output</span>} = <span class="q">'text'</span><span class="sc">;</span>
|
|
743 730 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">'comma'</span><span class="sc">;</span>
|
|
744 731 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">'yes'</span><span class="sc">;</span>
|
|
745 732
|
|
746 733 <span class="i">$Options</span>{<span class="w">size</span>} = <span class="n">1024</span><span class="sc">;</span>
|
|
747 734
|
|
748 735 <span class="i">$Options</span>{<span class="w">vectorstringformat</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
749 736
|
|
750 737 <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">"aromaticitymodel=s"</span><span class="cm">,</span> <span class="q">"atomidentifiertype|a=s"</span><span class="cm">,</span> <span class="q">"atomicinvariantstouse=s"</span><span class="cm">,</span> <span class="q">"functionalclassestouse=s"</span><span class="cm">,</span> <span class="q">"bitsorder=s"</span><span class="cm">,</span> <span class="q">"bitstringformat|b=s"</span><span class="cm">,</span> <span class="q">"compoundid=s"</span><span class="cm">,</span> <span class="q">"compoundidlabel=s"</span><span class="cm">,</span> <span class="q">"compoundidmode=s"</span><span class="cm">,</span> <span class="q">"datafields=s"</span><span class="cm">,</span> <span class="q">"datafieldsmode|d=s"</span><span class="cm">,</span> <span class="q">"filter|f=s"</span><span class="cm">,</span> <span class="q">"fingerprintslabel=s"</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"keeplargestcomponent|k=s"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"neighborhoodradius|n=s"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"output=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</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">"size|s=i"</span><span class="cm">,</span> <span class="q">"useperlcorerandom=s"</span><span class="cm">,</span> <span class="q">"vectorstringformat|v=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>
|
|
751 738 <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 $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
752 739 <span class="s">}</span>
|
|
753 740 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
754 741 <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>
|
|
755 742 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
756 743 <span class="s">}</span>
|
|
757 744 <span class="k">chdir</span> <span class="i">$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}: $! \n"</span><span class="sc">;</span>
|
|
758 745 <span class="s">}</span>
|
|
759 746 <span class="k">if</span> <span class="s">(</span>!<span class="i">Molecule::IsSupportedAromaticityModel</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">aromaticitymodel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
760 747 <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>
|
|
761 748 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{aromaticitymodel}, for option \"--AromaticityModel\" is not valid. Supported aromaticity models in current release of MayaChemTools: @SupportedModels\n"</span><span class="sc">;</span>
|
|
762 749 <span class="s">}</span>
|
|
763 750 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">atomidentifiertype</span>} !~ <span class="q">/^(AtomicInvariantsAtomTypes|FunctionalClassAtomTypes|DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
764 751 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, FunctionalClassAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n"</span><span class="sc">;</span>
|
|
765 752 <span class="s">}</span>
|
|
766 753 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">bitsorder</span>} !~ <span class="q">/^(Ascending|Descending)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
767 754 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{bitsorder}, for option \"--BitsOrder\" is not valid. Allowed values: Ascending or Descending\n"</span><span class="sc">;</span>
|
|
768 755 <span class="s">}</span>
|
|
769 756 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">bitstringformat</span>} !~ <span class="q">/^(BinaryString|HexadecimalString)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
770 757 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{bitstringformat}, for option \"-b, --bitstringformat\" is not valid. Allowed values: BinaryString or HexadecimalString\n"</span><span class="sc">;</span>
|
|
771 758 <span class="s">}</span>
|
|
772 759 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">compoundidmode</span>} !~ <span class="q">/^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
773 760 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{compoundidmode}, for option \"--CompoundIDMode\" is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n"</span><span class="sc">;</span>
|
|
774 761 <span class="s">}</span>
|
|
775 762 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsmode</span>} !~ <span class="q">/^(All|Common|Specify|CompoundID)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
776 763 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{datafieldsmode}, for option \"-d, --DataFieldsMode\" is not valid. Allowed values: All, Common, Specify or CompoundID\n"</span><span class="sc">;</span>
|
|
777 764 <span class="s">}</span>
|
|
778 765 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">filter</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
779 766 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{filter}, for option \"-f, --Filter\" is not valid. Allowed values: Yes or No\n"</span><span class="sc">;</span>
|
|
780 767 <span class="s">}</span>
|
|
781 768 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">keeplargestcomponent</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
782 769 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{keeplargestcomponent}, for option \"-k, --KeepLargestComponent\" is not valid. Allowed values: Yes or No\n"</span><span class="sc">;</span>
|
|
783 770 <span class="s">}</span>
|
|
784 771 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(ExtendedConnectivity|ExtendedConnectivityCount|ExtendedConnectivityBits)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
785 772 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: ExtendedConnectivity, ExtendedConnecticityCount, or ExtendedConnectivityBits\n"</span><span class="sc">;</span>
|
|
786 773 <span class="s">}</span>
|
|
787 774 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">neighborhoodradius</span>}<span class="s">)</span> && <span class="s">(</span><span class="i">$Options</span>{<span class="w">neighborhoodradius</span>} >= <span class="n">0</span><span class="s">)</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
788 775 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{neighborhoodradius}, for option \"-n, --NeighborhoodRadius\" is not valid. Allowed values: >= 0 \n"</span><span class="sc">;</span>
|
|
789 776 <span class="s">}</span>
|
|
790 777 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} !~ <span class="q">/^(SD|FP|text|all)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
791 778 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: SD, FP, text, or all\n"</span><span class="sc">;</span>
|
|
792 779 <span class="s">}</span>
|
|
793 780 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
794 781 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
795 782 <span class="s">}</span>
|
|
796 783 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
797 784 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: Yes or No\n"</span><span class="sc">;</span>
|
|
798 785 <span class="s">}</span>
|
|
799 786 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">size</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
800 787 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{size}, for option \"-s, --size\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
801 788 <span class="s">}</span>
|
|
802 789 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span> && <span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/^No$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
803 790 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not allowed with, semicolon value of \"--outdelim\" option: Fingerprints string use semicolon as delimiter for various data fields and must be quoted.\n"</span><span class="sc">;</span>
|
|
804 791 <span class="s">}</span>
|
|
805 792 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">useperlcorerandom</span>} !~ <span class="q">/^(Yes|No)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
806 793 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{useperlcorerandom}, for option \"--UsePerlCoreRandom\" is not valid. Allowed values: Yes or No\n"</span><span class="sc">;</span>
|
|
807 794 <span class="s">}</span>
|
|
808 795 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">vectorstringformat</span>} && <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>
|
|
809 796 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{vectorstringformat}, for option \"-v, --VectorStringFormat\" is not valid. Allowed values: ValuesString, IDsAndValuesString, IDsAndValuesPairsString, ValuesAndIDsString or ValuesAndIDsPairsString\n"</span><span class="sc">;</span>
|
|
810 797 <span class="s">}</span>
|
|
811 798 <span class="s">}</span>
|
|
812 799
|
|
813 <a name="EOF-"></a></pre>
|
|
814 <p> </p>
|
|
815 <br />
|
|
816 <center>
|
|
817 <img src="../../../images/h2o2.png">
|
|
818 </center>
|
|
819 </body>
|
|
820 </html>
|