annotate mayachemtools/bin/CalculatePhysicochemicalProperties.pl @ 9:ab29fa5c8c1f draft default tip

Uploaded
author deepakjadmin
date Thu, 15 Dec 2016 14:18:03 -0500
parents 73ae111cf86f
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1 #!/usr/bin/perl -w
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
2 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
3 # $RCSfile: CalculatePhysicochemicalProperties.pl,v $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
4 # $Date: 2015/02/28 20:46:19 $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
5 # $Revision: 1.20 $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
6 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
7 # Author: Manish Sud <msud@san.rr.com>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
8 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
9 # Copyright (C) 2015 Manish Sud. All rights reserved.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
10 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
11 # This file is part of MayaChemTools.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
12 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
13 # MayaChemTools is free software; you can redistribute it and/or modify it under
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
14 # the terms of the GNU Lesser General Public License as published by the Free
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
15 # Software Foundation; either version 3 of the License, or (at your option) any
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
16 # later version.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
17 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
18 # MayaChemTools is distributed in the hope that it will be useful, but without
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
19 # any warranty; without even the implied warranty of merchantability of fitness
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
20 # for a particular purpose. See the GNU Lesser General Public License for more
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
21 # details.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
22 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
23 # You should have received a copy of the GNU Lesser General Public License
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
26 # Boston, MA, 02111-1307, USA.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
27 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
28
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
29 use strict;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
30 use FindBin; use lib "$FindBin::Bin/../lib";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
31 use Getopt::Long;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
32 use File::Basename;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
33 use Text::ParseWords;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
34 use Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
35 use FileUtil;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
36 use TextUtil;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
37 use SDFileUtil;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
38 use MoleculeFileIO;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
39 use Molecule;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
40 use AtomTypes::AtomicInvariantsAtomTypes;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
41 use AtomTypes::FunctionalClassAtomTypes;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
42 use MolecularDescriptors::MolecularDescriptorsGenerator;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
43
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
44 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
45
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
46 # Autoflush STDOUT
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
47 $| = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
48
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
49 # Starting message...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
50 $ScriptName = basename($0);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
51 print "\n$ScriptName: Starting...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
52 $StartTime = new Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
53
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
54 # Get the options and setup script...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
55 SetupScriptUsage();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
56 if ($Options{help} || @ARGV < 1) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
57 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
58 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
59
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
60 my(@SDFilesList);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
61 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
62
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
63 # Process options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
64 print "Processing options...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
65 my(%OptionsInfo);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
66 ProcessOptions();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
67
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
68 # Setup information about input files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
69 print "Checking input SD file(s)...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
70 my(%SDFilesInfo);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
71 RetrieveSDFilesInfo();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
72
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
73 # Process input files..
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
74 my($FileIndex);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
75 if (@SDFilesList > 1) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
76 print "\nProcessing SD files...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
77 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
78 for $FileIndex (0 .. $#SDFilesList) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
79 if ($SDFilesInfo{FileOkay}[$FileIndex]) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
80 print "\nProcessing file $SDFilesList[$FileIndex]...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
81 CalculatePhysicochemicalProperties($FileIndex);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
82 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
83 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
84 print "\n$ScriptName:Done...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
85
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
86 $EndTime = new Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
87 $TotalTime = timediff ($EndTime, $StartTime);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
88 print "Total time: ", timestr($TotalTime), "\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
89
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
90 ###############################################################################
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
91
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
92 # Calculate physicochemical properties for a SD file...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
93 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
94 sub CalculatePhysicochemicalProperties {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
95 my($FileIndex) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
96 my($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount, $SDFile, $MoleculeFileIO, $Molecule, $MolecularDescriptorsGenerator, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
97
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
98 $SDFile = $SDFilesList[$FileIndex];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
99
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
100 # Setup output files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
101 $NewSDFileRef = ''; $NewTextFileRef = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
102 ($NewSDFileRef, $NewTextFileRef) = SetupAndOpenOutputFiles($FileIndex);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
103
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
104 # Setup molecular descriptor generator to calculate property values for specifed
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
105 # property names...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
106 $MolecularDescriptorsGenerator = SetupMolecularDescriptorsGenerator();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
107
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
108 ($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount) = ('0') x 4;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
109
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
110 $MoleculeFileIO = new MoleculeFileIO('Name' => $SDFile);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
111 $MoleculeFileIO->Open();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
112
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
113 COMPOUND: while ($Molecule = $MoleculeFileIO->ReadMolecule()) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
114 $CmpdCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
115
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
116 # Filter compound data before calculating physiochemical properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
117 if ($OptionsInfo{Filter}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
118 if (CheckAndFilterCompound($CmpdCount, $Molecule)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
119 $IgnoredCmpdCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
120 next COMPOUND;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
121 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
122 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
123
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
124 # Calculate properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
125 $PhysicochemicalPropertiesDataRef = CalculateMoleculeProperties($MolecularDescriptorsGenerator, $Molecule);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
126
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
127 if (!defined($PhysicochemicalPropertiesDataRef)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
128 $IgnoredCmpdCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
129 ProcessIgnoredCompound('PropertiesCalculationFailed', $CmpdCount, $Molecule);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
130 next COMPOUND;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
131 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
132
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
133 # Calculate any rule violations...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
134 if ($OptionsInfo{RuleOf5Violations} && $PhysicochemicalPropertiesDataRef->{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
135 $RuleOf5ViolationsCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
136 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
137
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
138 if ($OptionsInfo{RuleOf3Violations} && $PhysicochemicalPropertiesDataRef->{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
139 $RuleOf3ViolationsCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
140 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
141
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
142 # Write out calculate properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
143 WriteDataToOutputFiles($FileIndex, $CmpdCount, $Molecule, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
144 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
145 $MoleculeFileIO->Close();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
146
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
147 if ($OptionsInfo{SDOutput} && $NewSDFileRef) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
148 close $NewSDFileRef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
149 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
150 if ($OptionsInfo{TextOutput} && $NewTextFileRef) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
151 close $NewTextFileRef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
152 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
153
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
154 WriteCalculationSummaryStatistics($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
155 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
156
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
157 # Process compound being ignored due to problems in physicochemical properties calculation...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
158 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
159 sub ProcessIgnoredCompound {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
160 my($Mode, $CmpdCount, $Molecule) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
161 my($CmpdID, $DataFieldLabelAndValuesRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
162
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
163 $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
164 $CmpdID = SetupCmpdIDForTextFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
165
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
166 MODE: {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
167 if ($Mode =~ /^ContainsNonElementalData$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
168 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains atom data corresponding to non-elemental atom symbol(s)...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
169 next MODE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
170 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
171
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
172 if ($Mode =~ /^ContainsNoElementalData$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
173 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains no atom data...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
174 next MODE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
175 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
176
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
177 if ($Mode =~ /^PropertiesCalculationFailed$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
178 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Physicochemical properties calculation didn't succeed...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
179 next MODE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
180 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
181 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Physicochemical properties calculation didn't succeed...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
182 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
183 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
184
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
185 # Check and filter compounds....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
186 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
187 sub CheckAndFilterCompound {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
188 my($CmpdCount, $Molecule) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
189 my($ElementCount, $NonElementCount);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
190
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
191 ($ElementCount, $NonElementCount) = $Molecule->GetNumOfElementsAndNonElements();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
192
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
193 if ($NonElementCount) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
194 ProcessIgnoredCompound('ContainsNonElementalData', $CmpdCount, $Molecule);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
195 return 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
196 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
197
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
198 if (!$ElementCount) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
199 ProcessIgnoredCompound('ContainsNoElementalData', $CmpdCount, $Molecule);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
200 return 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
201 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
202
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
203 return 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
204 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
205
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
206 # Write out compounds physicochemical properties calculation summary statistics...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
207 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
208 sub WriteCalculationSummaryStatistics {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
209 my($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
210 my($ProcessedCmpdCount);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
211
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
212 $ProcessedCmpdCount = $CmpdCount - $IgnoredCmpdCount;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
213
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
214 print "\nNumber of compounds: $CmpdCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
215 print "Number of compounds processed successfully during physicochemical properties calculation: $ProcessedCmpdCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
216 print "Number of compounds ignored during physicochemical properties calculation: $IgnoredCmpdCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
217
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
218 if ($OptionsInfo{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
219 print "Number of compounds with one or more RuleOf5 violations: $RuleOf5ViolationsCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
220 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
221
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
222 if ($OptionsInfo{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
223 print "Number of compounds with one or more RuleOf3 violations: $RuleOf3ViolationsCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
224 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
225
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
226 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
227
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
228 # Open output files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
229 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
230 sub SetupAndOpenOutputFiles {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
231 my($FileIndex) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
232 my($NewSDFile, $NewTextFile, $NewSDFileRef, $NewTextFileRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
233
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
234 $NewSDFileRef = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
235 $NewTextFileRef = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
236
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
237 if ($OptionsInfo{SDOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
238 $NewSDFile = $SDFilesInfo{SDOutFileNames}[$FileIndex];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
239 print "Generating SD file $NewSDFile...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
240 open NEWSDFILE, ">$NewSDFile" or die "Error: Couldn't open $NewSDFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
241 $NewSDFileRef = \*NEWSDFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
242 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
243 if ($OptionsInfo{TextOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
244 $NewTextFile = $SDFilesInfo{TextOutFileNames}[$FileIndex];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
245 print "Generating text file $NewTextFile...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
246 open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
247 WriteTextFileCoulmnLabels($FileIndex, \*NEWTEXTFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
248 $NewTextFileRef = \*NEWTEXTFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
249 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
250 return ($NewSDFileRef, $NewTextFileRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
251 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
252
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
253 # Write calculated physicochemical properties and other data to appropriate output files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
254 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
255 sub WriteDataToOutputFiles {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
256 my($FileIndex, $CmpdCount, $Molecule, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
257 my($PropertyName, $PropertyValue);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
258
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
259 if ($OptionsInfo{SDOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
260 # Retrieve input compound string used to create molecule and write it out
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
261 # without last line containing a delimiter...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
262 my($CmpdString);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
263 $CmpdString = $Molecule->GetInputMoleculeString();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
264 $CmpdString =~ s/\$\$\$\$$//;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
265 print $NewSDFileRef "$CmpdString";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
266
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
267 # Write out calculated physicochemical properties data...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
268 for $PropertyName (@{$OptionsInfo{SpecifiedPropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
269 $PropertyValue = $PhysicochemicalPropertiesDataRef->{$PropertyName};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
270 print $NewSDFileRef "> <$PropertyName>\n$PropertyValue\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
271 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
272
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
273 # Write out RuleOf5 violations for molecule....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
274 if ($OptionsInfo{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
275 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf5Violations};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
276 print $NewSDFileRef "> <RuleOf5Violations>\n$PropertyValue\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
277 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
278
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
279 # Write out RuleOf3 violations for molecule....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
280 if ($OptionsInfo{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
281 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf3Violations};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
282 print $NewSDFileRef "> <RuleOf3Violations>\n$PropertyValue\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
283 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
284
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
285 # Write out delimiter...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
286 print $NewSDFileRef "\$\$\$\$\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
287 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
288
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
289 if ($OptionsInfo{TextOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
290 my($Line, $DataFieldLabelAndValuesRef, $DataFieldLabel, $DataFieldValue, @LineWords,);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
291
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
292 $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
293 @LineWords = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
294 if ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
295 push @LineWords, SetupCmpdIDForTextFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
296 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
297 elsif ($OptionsInfo{DataFieldsMode} =~ /^All$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
298 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
299 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
300 elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
301 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
302 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
303 elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
304 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$OptionsInfo{SpecifiedDataFields}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
305 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
306
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
307 # Append calculated physicochemical properties data...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
308 for $PropertyName (@{$OptionsInfo{SpecifiedPropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
309 $PropertyValue = $PhysicochemicalPropertiesDataRef->{$PropertyName};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
310 push @LineWords, $PropertyValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
311 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
312
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
313 # Write out RuleOf5 violations for molecule....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
314 if ($OptionsInfo{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
315 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf5Violations};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
316 push @LineWords, $PropertyValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
317 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
318
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
319 # Write out RuleOf3 violations for molecule....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
320 if ($OptionsInfo{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
321 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf3Violations};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
322 push @LineWords, $PropertyValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
323 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
324
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
325 $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
326 print $NewTextFileRef "$Line\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
327 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
328 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
329
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
330 # Write out approriate column labels to text file...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
331 sub WriteTextFileCoulmnLabels {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
332 my($FileIndex, $NewTextFileRef) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
333 my($Line, @LineWords);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
334
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
335 @LineWords = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
336 if ($OptionsInfo{DataFieldsMode} =~ /^All$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
337 push @LineWords, @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
338 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
339 elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
340 push @LineWords, @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
341 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
342 elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
343 push @LineWords, @{$OptionsInfo{SpecifiedDataFields}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
344 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
345 elsif ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
346 push @LineWords, $OptionsInfo{CompoundIDLabel};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
347 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
348 my($SpecifiedPropertyName);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
349
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
350 # Append physicochemical properties column labels...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
351 push @LineWords, @{$OptionsInfo{SpecifiedPropertyNames}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
352
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
353 # Write out RuleOf5 violations label...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
354 if ($OptionsInfo{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
355 push @LineWords, 'RuleOf5Violations';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
356 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
357
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
358 # Write out RuleOf3 violations label...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
359 if ($OptionsInfo{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
360 push @LineWords, 'RuleOf3Violations';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
361 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
362
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
363 $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
364 print $NewTextFileRef "$Line\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
365 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
366
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
367 # Generate compound ID for text files..
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
368 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
369 sub SetupCmpdIDForTextFiles {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
370 my($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
371 my($CmpdID);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
372
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
373 $CmpdID = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
374 if ($OptionsInfo{CompoundIDMode} =~ /^MolNameOrLabelPrefix$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
375 my($MolName);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
376 $MolName = $Molecule->GetName();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
377 $CmpdID = $MolName ? $MolName : "$OptionsInfo{CompoundID}${CmpdCount}";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
378 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
379 elsif ($OptionsInfo{CompoundIDMode} =~ /^LabelPrefix$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
380 $CmpdID = "$OptionsInfo{CompoundID}${CmpdCount}";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
381 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
382 elsif ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
383 my($SpecifiedDataField);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
384 $SpecifiedDataField = $OptionsInfo{CompoundID};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
385 $CmpdID = exists $DataFieldLabelAndValuesRef->{$SpecifiedDataField} ? $DataFieldLabelAndValuesRef->{$SpecifiedDataField} : '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
386 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
387 elsif ($OptionsInfo{CompoundIDMode} =~ /^MolName$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
388 $CmpdID = $Molecule->GetName();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
389 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
390 return $CmpdID;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
391 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
392
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
393 # Calculate physicochemical properties for molecule...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
394 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
395 sub CalculateMoleculeProperties {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
396 my($MolecularDescriptorsGenerator, $Molecule) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
397 my($PropertyName, $PropertyValue, $MolecularDescriptorsObject, %CalculatedPhysicochemicalProperties);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
398
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
399 %CalculatedPhysicochemicalProperties = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
400
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
401 if ($OptionsInfo{KeepLargestComponent}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
402 $Molecule->KeepLargestComponent();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
403 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
404
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
405 if (!$Molecule->DetectRings()) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
406 return undef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
407 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
408 $Molecule->SetAromaticityModel($OptionsInfo{AromaticityModel});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
409 $Molecule->DetectAromaticity();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
410
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
411 if ($OptionsInfo{AddHydrogens}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
412 $Molecule->AddHydrogens();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
413 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
414
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
415 # Calculate physicochemical properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
416 $MolecularDescriptorsGenerator->SetMolecule($Molecule);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
417 $MolecularDescriptorsGenerator->GenerateDescriptors();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
418
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
419 if (!$MolecularDescriptorsGenerator->IsDescriptorsGenerationSuccessful()) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
420 return undef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
421 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
422
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
423 %CalculatedPhysicochemicalProperties = $MolecularDescriptorsGenerator->GetDescriptorNamesAndValues();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
424
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
425 # Count RuleOf3 violations...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
426 if ($OptionsInfo{RuleOf3Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
427 CalculateRuleViolationsCount('RuleOf3Violations', \%CalculatedPhysicochemicalProperties);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
428 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
429
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
430 # Count RuleOf5 violations...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
431 if ($OptionsInfo{RuleOf5Violations}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
432 CalculateRuleViolationsCount('RuleOf5Violations', \%CalculatedPhysicochemicalProperties);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
433 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
434
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
435 return \%CalculatedPhysicochemicalProperties;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
436 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
437
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
438 # Setup molecular descriptor generator to calculate property values for specifed
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
439 # property names...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
440 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
441 sub SetupMolecularDescriptorsGenerator {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
442 my($PropertyName, $MolecularDescriptorsGenerator);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
443
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
444 $MolecularDescriptorsGenerator = new MolecularDescriptors::MolecularDescriptorsGenerator('Mode' => 'Specify', 'DescriptorNames' => \@{$OptionsInfo{SpecifiedPropertyNames}});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
445
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
446 # Setup molecular desciptor calculation parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
447 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('MolecularWeight')}) || exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('ExactMass')}) ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
448 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'WeightAndMassDescriptors', %{$OptionsInfo{PrecisionParametersMap}});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
449 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
450
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
451 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('RotatableBonds')})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
452 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'RotatableBondsDescriptors', %{$OptionsInfo{RotatableBondsParametersMap}});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
453 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
454
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
455 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('HydrogenBondDonors')}) || exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('HydrogenBondAcceptors')}) ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
456 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'HydrogenBondsDescriptors', 'HydrogenBondsType' => $OptionsInfo{HydrogenBonds});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
457 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
458
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
459 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('TPSA')})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
460 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'TPSADescriptors', %{$OptionsInfo{TPSAParametersMap}});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
461 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
462
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
463 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('MolecularComplexity')})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
464 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'MolecularComplexityDescriptors', %{$OptionsInfo{MolecularComplexityParametersMap}});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
465 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
466
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
467 return $MolecularDescriptorsGenerator;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
468 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
469
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
470 # Calculate RuleOf3 or RuleOf5 violations count...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
471 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
472 sub CalculateRuleViolationsCount {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
473 my($RuleViolationsType, $CalculatedPropertiesMapRef) = @_;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
474 my($RuleViolationsCount, $PropertyName);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
475
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
476 $RuleViolationsCount = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
477
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
478 RULEVIOLATIONSTYPE: {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
479 if ($RuleViolationsType =~ /^RuleOf3Violations$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
480 for $PropertyName (@{$OptionsInfo{RuleOf3PropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
481 if ($CalculatedPropertiesMapRef->{$PropertyName} > $OptionsInfo{RuleOf3MaxPropertyValuesMap}{$PropertyName}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
482 $RuleViolationsCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
483 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
484 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
485 last RULEVIOLATIONSTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
486 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
487
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
488 if ($RuleViolationsType =~ /^RuleOf5Violations$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
489 for $PropertyName (@{$OptionsInfo{RuleOf5PropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
490 if ($CalculatedPropertiesMapRef->{$PropertyName} > $OptionsInfo{RuleOf5MaxPropertyValuesMap}{$PropertyName}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
491 $RuleViolationsCount++;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
492 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
493 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
494 last RULEVIOLATIONSTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
495 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
496
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
497 die "Warning: Unknown rule violation type: $RuleViolationsType...";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
498 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
499
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
500 # Set rule violation count...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
501 $CalculatedPropertiesMapRef->{$RuleViolationsType} = $RuleViolationsCount;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
502
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
503 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
504
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
505 # Retrieve information about SD files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
506 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
507 sub RetrieveSDFilesInfo {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
508 my($SDFile, $Index, $FileDir, $FileExt, $FileName, $OutFileRoot, $TextOutFileExt, $SDOutFileExt, $NewSDFileName, $NewTextFileName, $CheckDataField, $CollectDataFields, $AllDataFieldsRef, $CommonDataFieldsRef);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
509
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
510 %SDFilesInfo = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
511 @{$SDFilesInfo{FileOkay}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
512 @{$SDFilesInfo{OutFileRoot}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
513 @{$SDFilesInfo{SDOutFileNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
514 @{$SDFilesInfo{TextOutFileNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
515 @{$SDFilesInfo{AllDataFieldsRef}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
516 @{$SDFilesInfo{CommonDataFieldsRef}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
517
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
518 $CheckDataField = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) && ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i)) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
519 $CollectDataFields = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^(All|Common)$/i)) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
520
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
521 FILELIST: for $Index (0 .. $#SDFilesList) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
522 $SDFile = $SDFilesList[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
523
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
524 $SDFilesInfo{FileOkay}[$Index] = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
525 $SDFilesInfo{OutFileRoot}[$Index] = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
526 $SDFilesInfo{SDOutFileNames}[$Index] = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
527 $SDFilesInfo{TextOutFileNames}[$Index] = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
528
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
529 $SDFile = $SDFilesList[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
530 if (!(-e $SDFile)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
531 warn "Warning: Ignoring file $SDFile: It doesn't exist\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
532 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
533 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
534 if (!CheckFileType($SDFile, "sd sdf")) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
535 warn "Warning: Ignoring file $SDFile: It's not a SD file\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
536 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
537 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
538
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
539 if ($CheckDataField) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
540 # Make sure data field exists in SD file..
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
541 my($CmpdString, $SpecifiedDataField, @CmpdLines, %DataFieldValues);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
542
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
543 @CmpdLines = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
544 open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
545 $CmpdString = ReadCmpdString(\*SDFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
546 close SDFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
547 @CmpdLines = split "\n", $CmpdString;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
548 %DataFieldValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
549 $SpecifiedDataField = $OptionsInfo{CompoundID};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
550 if (!exists $DataFieldValues{$SpecifiedDataField}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
551 warn "Warning: Ignoring file $SDFile: Data field value, $SpecifiedDataField, using \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\" doesn't exist\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
552 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
553 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
554 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
555
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
556 $AllDataFieldsRef = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
557 $CommonDataFieldsRef = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
558 if ($CollectDataFields) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
559 my($CmpdCount);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
560 open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
561 ($CmpdCount, $AllDataFieldsRef, $CommonDataFieldsRef) = GetAllAndCommonCmpdDataHeaderLabels(\*SDFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
562 close SDFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
563 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
564
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
565 # Setup output file names...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
566 $FileDir = ""; $FileName = ""; $FileExt = "";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
567 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
568
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
569 $TextOutFileExt = "csv";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
570 if ($Options{outdelim} =~ /^tab$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
571 $TextOutFileExt = "tsv";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
572 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
573 $SDOutFileExt = $FileExt;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
574
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
575 if ($OptionsInfo{OutFileRoot} && (@SDFilesList == 1)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
576 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($OptionsInfo{OutFileRoot});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
577 if ($RootFileName && $RootFileExt) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
578 $FileName = $RootFileName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
579 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
580 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
581 $FileName = $OptionsInfo{OutFileRoot};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
582 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
583 $OutFileRoot = $FileName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
584 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
585 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
586 $OutFileRoot = "${FileName}PhysicochemicalProperties";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
587 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
588
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
589 $NewSDFileName = "${OutFileRoot}.${SDOutFileExt}";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
590 $NewTextFileName = "${OutFileRoot}.${TextOutFileExt}";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
591
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
592 if ($OptionsInfo{SDOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
593 if ($SDFile =~ /$NewSDFileName/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
594 warn "Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
595 print "Specify a different name using \"-r --root\" option or use default name.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
596 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
597 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
598 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
599
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
600 if (!$OptionsInfo{OverwriteFiles}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
601 # Check SD and text outout files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
602 if ($OptionsInfo{SDOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
603 if (-e $NewSDFileName) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
604 warn "Warning: Ignoring file $SDFile: The file $NewSDFileName already exists\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
605 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
606 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
607 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
608 if ($OptionsInfo{TextOutput}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
609 if (-e $NewTextFileName) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
610 warn "Warning: Ignoring file $SDFile: The file $NewTextFileName already exists\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
611 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
612 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
613 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
614 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
615
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
616 $SDFilesInfo{FileOkay}[$Index] = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
617
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
618 $SDFilesInfo{OutFileRoot}[$Index] = $OutFileRoot;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
619 $SDFilesInfo{SDOutFileNames}[$Index] = $NewSDFileName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
620 $SDFilesInfo{TextOutFileNames}[$Index] = $NewTextFileName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
621
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
622 $SDFilesInfo{AllDataFieldsRef}[$Index] = $AllDataFieldsRef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
623 $SDFilesInfo{CommonDataFieldsRef}[$Index] = $CommonDataFieldsRef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
624 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
625 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
626
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
627 # Process option values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
628 sub ProcessOptions {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
629 %OptionsInfo = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
630
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
631 $OptionsInfo{AromaticityModel} = $Options{aromaticitymodel};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
632
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
633 # Process property name related options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
634 ProcessPropertyNamesOption();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
635
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
636 # Setup RuleOf3 and RuleOf5 violation calculations...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
637 $OptionsInfo{RuleOf3Violations} = ($Options{ruleof3violations} =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
638 $OptionsInfo{RuleOf5Violations} = ($Options{ruleof5violations} =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
639
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
640 $OptionsInfo{CompoundIDMode} = $Options{compoundidmode};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
641 $OptionsInfo{CompoundIDLabel} = $Options{compoundidlabel};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
642 $OptionsInfo{DataFieldsMode} = $Options{datafieldsmode};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
643
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
644 my(@SpecifiedDataFields);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
645 @SpecifiedDataFields = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
646
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
647 @{$OptionsInfo{SpecifiedDataFields}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
648 $OptionsInfo{CompoundID} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
649
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
650 if ($Options{datafieldsmode} =~ /^CompoundID$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
651 if ($Options{compoundidmode} =~ /^DataField$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
652 if (!$Options{compoundid}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
653 die "Error: You must specify a value for \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\". \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
654 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
655 $OptionsInfo{CompoundID} = $Options{compoundid};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
656 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
657 elsif ($Options{compoundidmode} =~ /^(LabelPrefix|MolNameOrLabelPrefix)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
658 $OptionsInfo{CompoundID} = $Options{compoundid} ? $Options{compoundid} : 'Cmpd';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
659 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
660 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
661 elsif ($Options{datafieldsmode} =~ /^Specify$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
662 if (!$Options{datafields}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
663 die "Error: You must specify a value for \"--DataFields\" option in \"Specify\" \"-d, --DataFieldsMode\". \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
664 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
665 @SpecifiedDataFields = split /\,/, $Options{datafields};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
666 push @{$OptionsInfo{SpecifiedDataFields}}, @SpecifiedDataFields;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
667 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
668
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
669 # Types of hydrogen bonds...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
670 $OptionsInfo{HydrogenBonds} = $Options{hydrogenbonds};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
671
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
672 # Process precision value parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
673 ProcessPrecisionOption();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
674
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
675 # Process rotatable bonds parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
676 ProcessRotatableBondsOption();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
677
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
678 # Process TPSA parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
679 ProcessTPSAOption();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
680
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
681 # Process molecular complexity parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
682 ProcessMolecularComplexityOption();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
683
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
684 $OptionsInfo{Filter} = ($Options{filter} =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
685
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
686 $OptionsInfo{KeepLargestComponent} = ($Options{keeplargestcomponent} =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
687
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
688 $OptionsInfo{Output} = $Options{output};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
689 $OptionsInfo{SDOutput} = ($Options{output} =~ /^(SD|Both)$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
690 $OptionsInfo{TextOutput} = ($Options{output} =~ /^(Text|Both)$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
691
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
692 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /tab/i ) ? "\t" : (($Options{outdelim} =~ /semicolon/i) ? "\;" : "\,");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
693 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
694
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
695 $OptionsInfo{OverwriteFiles} = $Options{overwrite} ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
696 $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
697 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
698
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
699 # Process property name related options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
700 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
701 sub ProcessPropertyNamesOption {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
702
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
703 # Setup supported physicochemical properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
704 my($SupportedProperty);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
705
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
706 @{$OptionsInfo{SupportedPropertyNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
707 %{$OptionsInfo{SupportedPropertyNamesMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
708
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
709 @{$OptionsInfo{RuleOf5PropertyNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
710 %{$OptionsInfo{RuleOf5MaxPropertyValuesMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
711
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
712 @{$OptionsInfo{RuleOf3PropertyNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
713 %{$OptionsInfo{RuleOf3MaxPropertyValuesMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
714
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
715 @{$OptionsInfo{DefaultPropertyNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
716
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
717 @{$OptionsInfo{SupportedPropertyNames}} = qw(MolecularWeight ExactMass HeavyAtoms Rings AromaticRings MolecularVolume RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP SMR TPSA Fsp3Carbons Sp3Carbons MolecularComplexity);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
718
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
719 @{$OptionsInfo{RuleOf5PropertyNames}} = qw(MolecularWeight HydrogenBondDonors HydrogenBondAcceptors SLogP);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
720 %{$OptionsInfo{RuleOf5MaxPropertyValuesMap}} = ('MolecularWeight' => 500, 'HydrogenBondDonors' => 5, 'HydrogenBondAcceptors' => 10, 'SLogP' => 5);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
721
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
722 @{$OptionsInfo{RuleOf3PropertyNames}} = qw(MolecularWeight RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP TPSA);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
723 %{$OptionsInfo{RuleOf3MaxPropertyValuesMap}} = ('MolecularWeight' => 300, 'RotatableBonds' => 3, 'HydrogenBondDonors' => 3, 'HydrogenBondAcceptors' => 3, 'SLogP' => 3, 'TPSA' => 60);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
724
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
725 @{$OptionsInfo{DefaultPropertyNames}} = qw(MolecularWeight HeavyAtoms MolecularVolume RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP TPSA);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
726
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
727 for $SupportedProperty (@{$OptionsInfo{SupportedPropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
728 $OptionsInfo{SupportedPropertyNamesMap}{lc($SupportedProperty)} = $SupportedProperty;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
729 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
730
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
731 # Process specified properties....
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
732 my($SpecifiedPropertyName, @SpecifiedPropertyNames, %SpecifiedPropertyNamesMap);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
733
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
734 @SpecifiedPropertyNames = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
735 %SpecifiedPropertyNamesMap = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
736
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
737 @{$OptionsInfo{SpecifiedPropertyNames}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
738 %{$OptionsInfo{SpecifiedPropertyNamesMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
739
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
740 if ($Options{mode} =~ /^All$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
741 @SpecifiedPropertyNames = @{$OptionsInfo{SupportedPropertyNames}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
742 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
743 elsif ($Options{mode} =~ /^RuleOf5$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
744 @SpecifiedPropertyNames = @{$OptionsInfo{RuleOf5PropertyNames}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
745 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
746 elsif ($Options{mode} =~ /^RuleOf3$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
747 @SpecifiedPropertyNames = @{$OptionsInfo{RuleOf3PropertyNames}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
748 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
749 elsif (IsEmpty($Options{mode})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
750 @SpecifiedPropertyNames = @{$OptionsInfo{DefaultPropertyNames}};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
751 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
752 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
753 # Comma delimited lisr of specified property names...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
754 my($Mode, $PropertyName, @PropertyNames, @UnsupportedPropertyNames);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
755
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
756 $Mode = $Options{mode};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
757 $Mode =~ s/ //g;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
758
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
759 @PropertyNames = split ",", $Mode;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
760 @UnsupportedPropertyNames = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
761
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
762 for $PropertyName (@PropertyNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
763 if (exists($OptionsInfo{SupportedPropertyNamesMap}{lc($PropertyName)})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
764 push @SpecifiedPropertyNames, $PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
765 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
766 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
767 push @UnsupportedPropertyNames, $PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
768 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
769 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
770 if (@UnsupportedPropertyNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
771 if (@UnsupportedPropertyNames > 1) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
772 warn "Error: The physicochemical property names specified - ", JoinWords(\@UnsupportedPropertyNames, ", ", 0)," - for option \"-m --mode\" are not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
773 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
774 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
775 warn "Error: The physicochemical property name specified, @UnsupportedPropertyNames , for option \"-m --mode\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
776 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
777 die "Allowed values:", JoinWords(\@{$OptionsInfo{SupportedPropertyNames}}, ", ", 0), "\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
778 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
779 if (!@SpecifiedPropertyNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
780 die "Error: No valid physicochemical property names specified for option \"-m --mode\".\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
781 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
782 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
783
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
784 # Set up specified property names map...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
785 PROPERTY: for $SpecifiedPropertyName (@SpecifiedPropertyNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
786 if (exists $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
787 warn "Warning: The physicochemical property name, $SpecifiedPropertyName, is specified multiple times as value of option \"-m --mode\" .\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
788 next PROPERTY;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
789 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
790 # Canonical specified property name...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
791 $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($SpecifiedPropertyName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
792 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
793
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
794 # Make sure for calculation of RuleOf3Violations, all appropriate property names are specified...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
795 if ($Options{ruleof3violations} =~ /^Yes$/i && $Options{mode} =~ /^RuleOf5$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
796 die "Error: The value specified, $Options{ruleof3violations}, for \"--RuleOf3Violations\" option in \"RuleOf5\" \"-m --Mode\" is not valid. You must specify RuleOf3 value for \"-m --Mode\" to calculate RuleOf3 violations.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
797 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
798
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
799 if ($Options{ruleof3violations} =~ /^Yes$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
800 my($RuleOf3PropertyName, @MissingRuleOf3Names);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
801
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
802 @MissingRuleOf3Names = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
803 PROPERTY: for $RuleOf3PropertyName (@{$OptionsInfo{RuleOf3PropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
804 if (exists $SpecifiedPropertyNamesMap{lc($RuleOf3PropertyName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
805 next PROPERTY;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
806 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
807 push @MissingRuleOf3Names, $RuleOf3PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
808
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
809 # Add property name to specified properties names list and map...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
810 push @SpecifiedPropertyNames, $RuleOf3PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
811 $SpecifiedPropertyNamesMap{lc($RuleOf3PropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($RuleOf3PropertyName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
812 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
813 if (@MissingRuleOf3Names) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
814 warn "Warning: The following physicochemical property names not specified in \"-m --Mode\" option are required for calculating RuleOf3Violations and have been added to the list of property names: @MissingRuleOf3Names\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
815 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
816 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
817
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
818 # Make sure for calculation of RuleOf5Violations, all appropriate property names are specified...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
819 if ($Options{ruleof5violations} =~ /^Yes$/i && $Options{mode} =~ /^RuleOf3$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
820 die "Error: The value specified, $Options{ruleof5violations}, for \"--RuleOf5Violations\" option in \"RuleOf3\" \"-m --Mode\" is not valid. You must specify RuleOf5 value for \"-m --Mode\" to calculate RuleOf5 violations.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
821 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
822
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
823 if ($Options{ruleof5violations} =~ /^Yes$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
824 my($RuleOf5PropertyName, @MissingRuleOf5Names);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
825
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
826 @MissingRuleOf5Names = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
827 PROPERTY: for $RuleOf5PropertyName (@{$OptionsInfo{RuleOf5PropertyNames}}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
828 if (exists $SpecifiedPropertyNamesMap{lc($RuleOf5PropertyName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
829 next PROPERTY;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
830 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
831 push @MissingRuleOf5Names, $RuleOf5PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
832
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
833 # Add property name to specified properties names list and map...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
834 push @SpecifiedPropertyNames, $RuleOf5PropertyName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
835 $SpecifiedPropertyNamesMap{lc($RuleOf5PropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($RuleOf5PropertyName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
836 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
837 if (@MissingRuleOf5Names) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
838 warn "Warning: The following physicochemical property names not specified in \"-m --Mode\" option are required for calculating RuleOf5Violations and have been added to the list of property names: @MissingRuleOf5Names\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
839 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
840 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
841 $OptionsInfo{Mode} = $Options{mode};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
842
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
843 # Setup canonical specified property names corresponding to supported names in mixed case...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
844 my(@SpecifiedCanonicalPropertyNames);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
845
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
846 @SpecifiedCanonicalPropertyNames = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
847 for $SpecifiedPropertyName (@SpecifiedPropertyNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
848 push @SpecifiedCanonicalPropertyNames, $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
849 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
850 @{$OptionsInfo{SpecifiedPropertyNames}} = @SpecifiedCanonicalPropertyNames;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
851 %{$OptionsInfo{SpecifiedPropertyNamesMap}} = %SpecifiedPropertyNamesMap;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
852
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
853 # Based on specified property names, figure out whether hydrogens need to be added before
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
854 # calculation of properties...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
855 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
856 $OptionsInfo{AddHydrogens} = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
857 if (exists($SpecifiedPropertyNamesMap{lc('MolecularVolume')}) || exists($SpecifiedPropertyNamesMap{lc('SLogP')}) || exists($SpecifiedPropertyNamesMap{lc('SMR')})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
858 $OptionsInfo{AddHydrogens} = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
859 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
860 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
861
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
862 # Process precision option...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
863 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
864 sub ProcessPrecisionOption {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
865 my($ParameterName, $ParameterValue, %PrecisionParametersMap, %PrecisionParameterNamesMap);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
866
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
867 %{$OptionsInfo{PrecisionParametersMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
868
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
869 %PrecisionParametersMap = ('WeightPrecision' => 2, 'MassPrecision' => 4);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
870 %PrecisionParameterNamesMap = ('molecularweight' => 'WeightPrecision', 'exactmass' => 'MassPrecision');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
871
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
872 if ($Options{precision}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
873 # Process specified values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
874 my($Index, $SpecifiedPrecision, @SpecifiedPrecisionValuePairs);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
875
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
876 $SpecifiedPrecision = $Options{precision};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
877 $SpecifiedPrecision =~ s/ //g;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
878 @SpecifiedPrecisionValuePairs = split ",", $SpecifiedPrecision;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
879 if (@SpecifiedPrecisionValuePairs % 2) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
880 die "Error: Invalid number of values specified using \"--Precision\" option: It must contain even number of values.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
881 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
882 for ($Index = 0; (($Index + 1) < @SpecifiedPrecisionValuePairs); $Index += 2 ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
883 $ParameterName = $SpecifiedPrecisionValuePairs[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
884 $ParameterValue = $SpecifiedPrecisionValuePairs[$Index + 1];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
885 if (!exists $PrecisionParameterNamesMap{lc($ParameterName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
886 die "Error: The precision parameter name specified, $ParameterName, for option \"--Precision\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
887 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
888 if (!IsPositiveInteger($ParameterValue)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
889 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--Precision\" is not valid. Allowed values: positive integer. \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
890 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
891 $ParameterName = $PrecisionParameterNamesMap{lc($ParameterName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
892 $PrecisionParametersMap{$ParameterName} = $ParameterValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
893 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
894 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
895 $OptionsInfo{Precision} = $Options{precision};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
896 %{$OptionsInfo{PrecisionParametersMap}} = %PrecisionParametersMap;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
897 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
898
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
899 # Process rotatable bonds option...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
900 sub ProcessRotatableBondsOption {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
901 my($ParameterName, $ParameterValue, %RotatableBondsParametersMap, %RotatableBondsParameterNamesMap);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
902
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
903 %{$OptionsInfo{RotatableBondsParametersMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
904 %RotatableBondsParametersMap = ('IgnoreTerminalBonds' => 1, 'IgnoreBondsToTripleBonds' => 1, 'IgnoreAmideBonds' => 1, 'IgnoreThioamideBonds' => 1, 'IgnoreSulfonamideBonds' => 1);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
905
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
906 for $ParameterName (keys %RotatableBondsParametersMap) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
907 $RotatableBondsParameterNamesMap{lc($ParameterName)} = $ParameterName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
908 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
909
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
910 if ($Options{rotatablebonds}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
911 # Process specified values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
912 my($Index, $SpecifiedRotatableBonds, @SpecifiedRotatableBondsValuePairs);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
913
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
914 $SpecifiedRotatableBonds = $Options{rotatablebonds};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
915 $SpecifiedRotatableBonds =~ s/ //g;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
916 @SpecifiedRotatableBondsValuePairs = split ",", $SpecifiedRotatableBonds;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
917 if (@SpecifiedRotatableBondsValuePairs % 2) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
918 die "Error: Invalid number of values specified using \"--RotatableBonds\" option: It must contain even number of values.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
919 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
920 for ($Index = 0; (($Index + 1) < @SpecifiedRotatableBondsValuePairs); $Index += 2 ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
921 $ParameterName = $SpecifiedRotatableBondsValuePairs[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
922 $ParameterValue = $SpecifiedRotatableBondsValuePairs[$Index + 1];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
923 if (!exists $RotatableBondsParameterNamesMap{lc($ParameterName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
924 die "Error: The rotatable bonds parameter name specified, $ParameterName, for option \"--RotatableBonds\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
925 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
926 if ($ParameterValue !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
927 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--RotatableBonds\" is not valid. Allowed values: Yes or No. \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
928 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
929 $ParameterName = $RotatableBondsParameterNamesMap{lc($ParameterName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
930 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
931 $RotatableBondsParametersMap{$ParameterName} = $ParameterValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
932 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
933 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
934 $OptionsInfo{RotatableBonds} = $Options{rotatablebonds};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
935 %{$OptionsInfo{RotatableBondsParametersMap}} = %RotatableBondsParametersMap;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
936 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
937
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
938 # Process TPSA option...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
939 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
940 sub ProcessTPSAOption {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
941 my($ParameterName, $ParameterValue, %TPSAParametersMap, %TPSAParameterNamesMap);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
942
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
943 %{$OptionsInfo{TPSAParametersMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
944
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
945 %TPSAParametersMap = ('IgnorePhosphorus' => 1, 'IgnoreSulfur' => 1);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
946 for $ParameterName (keys %TPSAParametersMap) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
947 $TPSAParameterNamesMap{lc($ParameterName)} = $ParameterName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
948 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
949
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
950 if ($Options{tpsa}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
951 # Process specified values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
952 my($Index, $SpecifiedTPSA, @SpecifiedTPSAValuePairs);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
953
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
954 $SpecifiedTPSA = $Options{tpsa};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
955 $SpecifiedTPSA =~ s/ //g;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
956 @SpecifiedTPSAValuePairs = split ",", $SpecifiedTPSA;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
957 if (@SpecifiedTPSAValuePairs % 2) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
958 die "Error: Invalid number of values specified using \"--TPSA\" option: It must contain even number of values.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
959 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
960 for ($Index = 0; (($Index + 1) < @SpecifiedTPSAValuePairs); $Index += 2 ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
961 $ParameterName = $SpecifiedTPSAValuePairs[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
962 $ParameterValue = $SpecifiedTPSAValuePairs[$Index + 1];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
963 if (!exists $TPSAParameterNamesMap{lc($ParameterName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
964 die "Error: The TPSA parameter name specified, $ParameterName, for option \"--TPSA\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
965 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
966 if ($ParameterValue !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
967 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--TPSA\" is not valid. Allowed values: Yes or No. \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
968 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
969 $ParameterName = $TPSAParameterNamesMap{lc($ParameterName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
970 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
971 $TPSAParametersMap{$ParameterName} = $ParameterValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
972 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
973 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
974 $OptionsInfo{TPSA} = $Options{tpsa};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
975 %{$OptionsInfo{TPSAParametersMap}} = %TPSAParametersMap;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
976 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
977
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
978 # Process molecular complexity parameters...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
979 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
980 sub ProcessMolecularComplexityOption {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
981 my($MolecularComplexityType, $ParameterName, $ParameterValue, @ParameterNames, @ParameterValues, @AtomIdentifierTypeParameters, %ComplexityParametersMap, %ComplexityParameterNamesMap);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
982
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
983 %{$OptionsInfo{MolecularComplexityParametersMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
984
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
985 %ComplexityParametersMap = ('MolecularComplexityType' => '', 'AtomIdentifierType' => '',
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
986 'AtomicInvariantsToUse' => '', 'FunctionalClassesToUse' => '',
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
987 'MACCSKeysSize' => '166', 'NeighborhoodRadius' => '2',
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
988 'MinPathLength' => '1', 'MaxPathLength' => '8', 'UseBondSymbols' => '1',
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
989 'MinDistance' => '1', 'MaxDistance' => '10', 'UseTriangleInequality' => '',
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
990 'DistanceBinSize' => '2', 'NormalizationMethodology' => 'None');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
991
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
992 %ComplexityParameterNamesMap = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
993 for $ParameterName (keys %ComplexityParametersMap) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
994 $ComplexityParameterNamesMap{lc($ParameterName)} = $ParameterName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
995 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
996
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
997 if ($Options{molecularcomplexity}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
998 # Process specified values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
999 my($Index, $SpecifiedComplexity, @SpecifiedComplexityValuePairs);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1000
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1001 $SpecifiedComplexity = $Options{molecularcomplexity};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1002
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1003 @SpecifiedComplexityValuePairs = split ",", $SpecifiedComplexity;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1004 if (@SpecifiedComplexityValuePairs % 2) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1005 die "Error: Invalid number of values specified using \"--MolecularComplexity\" option: It must contain even number of values.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1006 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1007
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1008 for ($Index = 0; (($Index + 1) < @SpecifiedComplexityValuePairs); $Index += 2 ) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1009 $ParameterName = $SpecifiedComplexityValuePairs[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1010 $ParameterValue = $SpecifiedComplexityValuePairs[$Index + 1];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1011
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1012 $ParameterName = RemoveLeadingAndTrailingWhiteSpaces($ParameterName);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1013 $ParameterValue = RemoveLeadingAndTrailingWhiteSpaces($ParameterValue);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1014
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1015 if (!exists $ComplexityParameterNamesMap{lc($ParameterName)}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1016 die "Error: The molecular complexity parameter name specified, $ParameterName, for option \"--MolecularComplexity\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1017 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1018 $ParameterName = $ComplexityParameterNamesMap{lc($ParameterName)};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1019
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1020 if ($ParameterName =~ /^AtomicInvariantsToUse$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1021 my($AtomSymbolFound);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1022
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1023 $AtomSymbolFound = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1024 @ParameterValues = split(' ', $ParameterValue);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1025 for $ParameterValue (@ParameterValues) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1026 if (!AtomTypes::AtomicInvariantsAtomTypes::IsAtomicInvariantAvailable($ParameterValue)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1027 die "Error: The atomic invariant specified, $ParameterValue, for AtomicInvariantsToUse in option \"--MolecularComplexity\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1028 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1029 if ($ParameterValue =~ /^(AS|AtomSymbol)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1030 $AtomSymbolFound = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1031 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1032 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1033 if (!$AtomSymbolFound) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1034 die "Error: The atomic invariants specified using AtomicInvariantsToUse in option \"--MolecularComplexity\" is not valid: AtomicInvariant atom symbol, AS or AtomSymbol, must be specified.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1035 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1036 $ParameterValue = JoinWords(\@ParameterValues, ",", 0);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1037 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1038 elsif ($ParameterName =~ /^FunctionalClassesToUse$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1039 @ParameterValues = split(' ', $ParameterValue);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1040 for $ParameterValue (@ParameterValues) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1041 if (!AtomTypes::FunctionalClassAtomTypes::IsFunctionalClassAvailable($ParameterValue)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1042 die "Error: The functional class specified, $ParameterValue, for FunctionalClassesToUse in option \"--MolecularComplexity\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1043 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1044 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1045 $ParameterValue = JoinWords(\@ParameterValues, ",", 0);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1046 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1047 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1048 if ($ParameterValue =~ / /) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1049 $ParameterValue =~ s/ //g;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1050 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1051 if ($ParameterValue =~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1052 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1053 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1054 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1055
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1056 if ($ParameterName =~ /^MolecularComplexityType$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1057 if ($ParameterValue !~ /^(AtomTypesFingerprints|ExtendedConnectivityFingerprints|MACCSKeys|PathLengthFingerprints|TopologicalAtomPairsFingerprints|TopologicalAtomTripletsFingerprints|TopologicalAtomTorsionsFingerprints|TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1058 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: AtomTypesFingerprints, ExtendedConnectivityFingerprints, MACCSKeys, PathLengthFingerprints, TopologicalAtomPairsFingerprints, TopologicalAtomTripletsFingerprints, TopologicalAtomTorsionsFingerprints, TopologicalPharmacophoreAtomPairsFingerprints, or TopologicalPharmacophoreAtomTripletsFingerprints..\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1059 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1060 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1061 elsif ($ParameterName =~ /^AtomIdentifierType$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1062 if ($ParameterValue !~ /^(AtomicInvariantsAtomTypes|FunctionalClassAtomTypes|DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1063 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, FunctionalClassAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes and UFFAtomTypes.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1064 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1065 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1066 elsif ($ParameterName =~ /^(MACCSKeysSize|MinPathLength|MaxPathLength|MinDistance|MaxDistance|DistanceBinSize)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1067 if (!IsPositiveInteger($ParameterValue)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1068 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: positive integer. \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1069 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1070 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1071 elsif ($ParameterName =~ /^NeighborhoodRadius$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1072 if (!(IsInteger($ParameterValue) && $ParameterValue >=0)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1073 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: 0 or positive integer. \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1074 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1075 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1076 elsif ($ParameterName =~ /^NormalizationMethodology$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1077 if ($ParameterValue !~ /^(None|ByHeavyAtomsCount|ByPossibleKeysCount)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1078 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: None, ByHeavyAtomsCount, or ByPossibleKeysCount\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1079 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1080 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1081 $ComplexityParametersMap{$ParameterName} = $ParameterValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1082 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1083
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1084 if ($ComplexityParametersMap{MACCSKeysSize} !~ /^(166|322)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1085 die "Error: The parameter value specified, $ComplexityParametersMap{MACCSKeysSize}, for parameter name, MACCSKeysSize in option \"--MolecularComplexity\" is not valid. Allowed values: 166 or 322\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1086 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1087 if ($ComplexityParametersMap{MinPathLength} > $ComplexityParametersMap{MaxPathLength}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1088 die "Error: The parameter value specified for MinPathLength, $ComplexityParametersMap{MinPathLength}, must be <= MaxPathLength, $ComplexityParametersMap{MaxPathLength} ...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1089 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1090 if ($ComplexityParametersMap{MinDistance} > $ComplexityParametersMap{MaxDistance}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1091 die "Error: The parameter value specified for MinDistance, $ComplexityParametersMap{MinDistance}, must be <= MaxDistance, $ComplexityParametersMap{MaxDistance} ...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1092 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1093 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1094
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1095 # Set default parameter values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1096
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1097 if (IsEmpty($ComplexityParametersMap{MolecularComplexityType})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1098 $ComplexityParametersMap{MolecularComplexityType} = 'MACCSKeys';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1099 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1100 $MolecularComplexityType = $ComplexityParametersMap{MolecularComplexityType};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1101
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1102
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1103 if (IsEmpty($ComplexityParametersMap{AtomIdentifierType})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1104 $ComplexityParametersMap{AtomIdentifierType} = ($MolecularComplexityType =~ /^(TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) ? "FunctionalClassAtomTypes" : "AtomicInvariantsAtomTypes";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1105 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1106
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1107 if (IsEmpty($ComplexityParametersMap{AtomicInvariantsToUse})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1108 my($AtomicInvariantsToUse);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1109
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1110 if ($MolecularComplexityType =~ /^(AtomTypesFingerprints|TopologicalAtomPairsFingerprints|TopologicalAtomTripletsFingerprints|TopologicalAtomTorsionsFingerprints)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1111 $AtomicInvariantsToUse = "AS,X,BO,H,FC";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1112 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1113 elsif ($MolecularComplexityType =~ /^ExtendedConnectivityFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1114 $AtomicInvariantsToUse = "AS,X,BO,H,FC,MN";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1115 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1116 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1117 $AtomicInvariantsToUse = "AS";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1118 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1119 $ComplexityParametersMap{AtomicInvariantsToUse} = $AtomicInvariantsToUse;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1120 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1121
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1122 if (IsEmpty($ComplexityParametersMap{FunctionalClassesToUse})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1123 my($FunctionalClassesToUse);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1124
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1125 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomPairsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1126 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1127 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1128 elsif ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1129 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H,Ar";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1130 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1131 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1132 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H,Ar,Hal";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1133 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1134 $ComplexityParametersMap{FunctionalClassesToUse} = $FunctionalClassesToUse;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1135 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1136
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1137 my(@AtomicInvariantsToUse);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1138 @AtomicInvariantsToUse = split ',', $ComplexityParametersMap{AtomicInvariantsToUse};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1139 $ComplexityParametersMap{AtomicInvariantsToUse} = \@AtomicInvariantsToUse;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1140
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1141 my(@FunctionalClassesToUse);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1142 @FunctionalClassesToUse = split ',', $ComplexityParametersMap{FunctionalClassesToUse};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1143 $ComplexityParametersMap{FunctionalClassesToUse} = \@FunctionalClassesToUse;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1144
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1145 if (IsEmpty($ComplexityParametersMap{UseTriangleInequality})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1146 $ComplexityParametersMap{UseTriangleInequality} = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1147 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1148 $ComplexityParametersMap{UseTriangleInequality} = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1149 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1150 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1151
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1152 if ($MolecularComplexityType =~ /^(TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1153 if ($ComplexityParametersMap{AtomIdentifierType} !~ /^FunctionalClassAtomTypes$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1154 die "Error: The parameter value specified for AtomIdentifierType, $ComplexityParametersMap{AtomIdentifierType}, in option \"--MolecularComplexity\" is not valid for MolecularComplexityType, $MolecularComplexityType: Allowed value: FunctionalClassAtomTypes...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1155 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1156 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1157
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1158 # Set up approprate paremeter names for specified molecular complexity...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1159
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1160 @ParameterNames = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1161 push @ParameterNames, 'MolecularComplexityType';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1162
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1163 @AtomIdentifierTypeParameters = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1164 push @AtomIdentifierTypeParameters, 'AtomIdentifierType';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1165 if ($ComplexityParametersMap{AtomIdentifierType} =~ /^AtomicInvariantsAtomTypes$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1166 push @AtomIdentifierTypeParameters, 'AtomicInvariantsToUse';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1167 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1168 elsif ($ComplexityParametersMap{AtomIdentifierType} =~ /^FunctionalClassAtomTypes$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1169 push @AtomIdentifierTypeParameters, 'FunctionalClassesToUse';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1170 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1171
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1172 COMPLEXITYTYPE: {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1173 if ($MolecularComplexityType =~ /^AtomTypesFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1174 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1175 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1176 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1177 if ($MolecularComplexityType =~ /^ExtendedConnectivityFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1178 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1179 push @ParameterNames, ('NeighborhoodRadius', 'NormalizationMethodology');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1180 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1181 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1182 if ($MolecularComplexityType =~ /^MACCSKeys$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1183 push @ParameterNames, 'MACCSKeysSize';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1184 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1185 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1186 if ($MolecularComplexityType =~ /^PathLengthFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1187 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1188 push @ParameterNames, ('MinPathLength', 'MaxPathLength', 'UseBondSymbols');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1189 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1190 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1191 if ($MolecularComplexityType =~ /^TopologicalAtomPairsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1192 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1193 push @ParameterNames, ('MinDistance', 'MaxDistance');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1194 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1195 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1196 if ($MolecularComplexityType =~ /^TopologicalAtomTripletsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1197 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1198 push @ParameterNames, ('MinDistance', 'MaxDistance', 'UseTriangleInequality');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1199 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1200 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1201 if ($MolecularComplexityType =~ /^TopologicalAtomTorsionsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1202 push @ParameterNames, @AtomIdentifierTypeParameters;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1203 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1204 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1205 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomPairsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1206 push @ParameterNames, ('AtomIdentifierType', 'FunctionalClassesToUse', 'MinDistance', 'MaxDistance', 'NormalizationMethodology');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1207 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1208 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1209 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1210 push @ParameterNames, ('AtomIdentifierType', 'FunctionalClassesToUse', 'MinDistance', 'MaxDistance', 'UseTriangleInequality', 'NormalizationMethodology', 'DistanceBinSize');
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1211 last COMPLEXITYTYPE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1212 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1213 die "Error: The parameter value specified, $ParameterValue, for parameter name MolecularComplexityType using \"--MolecularComplexity\" is not valid.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1214 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1215
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1216 $OptionsInfo{MolecularComplexity} = $Options{molecularcomplexity};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1217
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1218 %{$OptionsInfo{MolecularComplexityParametersMap}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1219 for $ParameterName (@ParameterNames) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1220 $ParameterValue = $ComplexityParametersMap{$ParameterName};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1221 $OptionsInfo{MolecularComplexityParametersMap}{$ParameterName} = $ParameterValue;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1222 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1223 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1224
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1225 # Setup script usage and retrieve command line arguments specified using various options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1226 sub SetupScriptUsage {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1227
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1228 # Retrieve all the options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1229 %Options = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1230
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1231 $Options{aromaticitymodel} = 'MayaChemToolsAromaticityModel';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1232
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1233 $Options{compoundidmode} = 'LabelPrefix';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1234 $Options{compoundidlabel} = 'CompoundID';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1235 $Options{datafieldsmode} = 'CompoundID';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1236
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1237 $Options{filter} = 'Yes';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1238
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1239 $Options{hydrogenbonds} = 'HBondsType2';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1240
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1241 $Options{keeplargestcomponent} = 'Yes';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1242
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1243 # Default mode values are set later...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1244 $Options{mode} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1245
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1246 # Default moelcular complexity values are set later...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1247 $Options{molecularcomplexity} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1248
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1249 # Default precision values are set later...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1250 $Options{precision} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1251
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1252 $Options{output} = 'text';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1253 $Options{outdelim} = 'comma';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1254 $Options{quote} = 'yes';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1255
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1256 # Default rotatable bond parameter values are set later...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1257 $Options{rotatablebonds} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1258
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1259 $Options{ruleof3violations} = 'No';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1260 $Options{ruleof5violations} = 'No';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1261
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1262 # Default TPSA paramater values are set later...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1263 $Options{tpsa} = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1264
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1265 if (!GetOptions(\%Options, "aromaticitymodel=s", "compoundid=s", "compoundidlabel=s", "compoundidmode=s", "datafields=s", "datafieldsmode|d=s", "filter|f=s", "help|h", "hydrogenbonds=s", "keeplargestcomponent|k=s", "mode|m=s", "molecularcomplexity=s", "outdelim=s", "output=s", "overwrite|o", "precision=s", "rotatablebonds=s", "ruleof3violations=s", "ruleof5violations=s", "quote|q=s", "root|r=s", "tpsa=s", "workingdir|w=s")) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1266 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1267 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1268 if ($Options{workingdir}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1269 if (! -d $Options{workingdir}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1270 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1271 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1272 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1273 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1274 if (!Molecule::IsSupportedAromaticityModel($Options{aromaticitymodel})) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1275 my(@SupportedModels) = Molecule::GetSupportedAromaticityModels();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1276 die "Error: The value specified, $Options{aromaticitymodel}, for option \"--AromaticityModel\" is not valid. Supported aromaticity models in current release of MayaChemTools: @SupportedModels\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1277 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1278 if ($Options{compoundidmode} !~ /^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1279 die "Error: The value specified, $Options{compoundidmode}, for option \"--CompoundIDMode\" is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1280 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1281 if ($Options{datafieldsmode} !~ /^(All|Common|Specify|CompoundID)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1282 die "Error: The value specified, $Options{datafieldsmode}, for option \"-d, --DataFieldsMode\" is not valid. Allowed values: All, Common, Specify or CompoundID\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1283 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1284 if ($Options{filter} !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1285 die "Error: The value specified, $Options{filter}, for option \"-f, --Filter\" is not valid. Allowed values: Yes or No\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1286 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1287 if ($Options{hydrogenbonds} !~ /^(HBondsType1|HydrogenBondsType1|HBondsType2|HydrogenBondsType2)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1288 die "Error: The value specified, $Options{hydrogenbonds}, for option \"--HydrogenBonds\" is not valid. Allowed values: HBondsType1, HydrogenBondsType1, HBondsType2, HydrogenBondsType2\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1289 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1290 if ($Options{keeplargestcomponent} !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1291 die "Error: The value specified, $Options{keeplargestcomponent}, for option \"-k, --KeepLargestComponent\" is not valid. Allowed values: Yes or No\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1292 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1293 if ($Options{output} !~ /^(SD|text|both)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1294 die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: SD, text, or both\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1295 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1296 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1297 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1298 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1299 if ($Options{quote} !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1300 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: Yes or No\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1301 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1302 if ($Options{ruleof3violations} !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1303 die "Error: The value specified, $Options{ruleof3violations}, for option \"--RuleOf3Violations\" is not valid. Allowed values: Yes or No\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1304 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1305 if ($Options{ruleof5violations} !~ /^(Yes|No)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1306 die "Error: The value specified, $Options{ruleof5violations}, for option \"--RuleOf5Violations\" is not valid. Allowed values: Yes or No\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1307 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1308 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1309
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1310 __END__
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1311
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1312 =head1 NAME
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1313
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1314 CalculatePhysicochemicalProperties.pl - Calculate physicochemical properties for SD files
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1315
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1316 =head1 SYNOPSIS
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1317
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1318 CalculatePhysicochemicalProperties.pl SDFile(s)...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1319
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1320 PhysicochemicalProperties.pl [B<--AromaticityModel> I<AromaticityModelType>]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1321 [B<--CompoundID> DataFieldName or LabelPrefixString]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1322 [B<--CompoundIDLabel> text] [B<--CompoundIDMode>] [B<--DataFields> "FieldLabel1, FieldLabel2,..."]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1323 [B<-d, --DataFieldsMode> All | Common | Specify | CompoundID] [B<-f, --Filter> Yes | No] [B<-h, --help>]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1324 [B<--HydrogenBonds> HBondsType1 | HBondsType2] [B<-k, --KeepLargestComponent> Yes | No]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1325 [B<-m, --mode> All | RuleOf5 | RuleOf3 | "name1, [name2,...]"]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1326 [B<--MolecularComplexity> I<Name,Value, [Name,Value,...]>]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1327 [B<--OutDelim> comma | tab | semicolon] [B<--output> SD | text | both] [B<-o, --overwrite>]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1328 [B<--Precision> Name,Number,[Name,Number,..]] [B<--RotatableBonds> Name,Value, [Name,Value,...]]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1329 [B<--RuleOf3Violations> Yes | No] [B<--RuleOf5Violations> Yes | No]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1330 [B<-q, --quote> Yes | No] [B<-r, --root> RootName]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1331 [B<-w, --WorkingDir> dirname] SDFile(s)...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1332
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1333 =head1 DESCRIPTION
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1334
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1335 Calculate physicochemical properties for I<SDFile(s)> and create appropriate SD or CSV/TSV
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1336 text file(s) containing calculated properties.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1337
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1338 The current release of MayaChemTools supports the calculation of these physicochemical
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1339 properties:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1340
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1341 MolecularWeight, ExactMass, HeavyAtoms, Rings, AromaticRings,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1342 van der Waals MolecularVolume [ Ref 93 ], RotatableBonds,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1343 HydrogenBondDonors, HydrogenBondAcceptors, LogP and
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1344 Molar Refractivity (SLogP and SMR) [ Ref 89 ], Topological Polar
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1345 Surface Area (TPSA) [ Ref 90 ], Fraction of SP3 carbons (Fsp3Carbons)
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1346 and SP3 carbons (Sp3Carbons) [ Ref 115-116, Ref 119 ],
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1347 MolecularComplexity [ Ref 117-119 ]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1348
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1349 Multiple SDFile names are separated by spaces. The valid file extensions are I<.sdf>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1350 and I<.sd>. All other file names are ignored. All the SD files in a current directory
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1351 can be specified either by I<*.sdf> or the current directory name.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1352
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1353 The calculation of molecular complexity using I<MolecularComplexityType> parameter
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1354 corresponds to the number of bits-set or unique keys [ Ref 117-119 ] in molecular fingerprints.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1355 Default value for I<MolecularComplexityType>: I<MACCSKeys> of size 166. The calculation
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1356 of MACCSKeys is relatively expensive and can take rather substantial amount of time.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1357
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1358 =head1 OPTIONS
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1359
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1360 =over 4
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1361
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1362 =item B<--AromaticityModel> I<MDLAromaticityModel | TriposAromaticityModel | MMFFAromaticityModel | ChemAxonBasicAromaticityModel | ChemAxonGeneralAromaticityModel | DaylightAromaticityModel | MayaChemToolsAromaticityModel>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1363
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1364 Specify aromaticity model to use during detection of aromaticity. Possible values in the current
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1365 release are: I<MDLAromaticityModel, TriposAromaticityModel, MMFFAromaticityModel,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1366 ChemAxonBasicAromaticityModel, ChemAxonGeneralAromaticityModel, DaylightAromaticityModel
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1367 or MayaChemToolsAromaticityModel>. Default value: I<MayaChemToolsAromaticityModel>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1368
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1369 The supported aromaticity model names along with model specific control parameters
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1370 are defined in B<AromaticityModelsData.csv>, which is distributed with the current release
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1371 and is available under B<lib/data> directory. B<Molecule.pm> module retrieves data from
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1372 this file during class instantiation and makes it available to method B<DetectAromaticity>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1373 for detecting aromaticity corresponding to a specific model.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1374
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1375 =item B<--CompoundID> I<DataFieldName or LabelPrefixString>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1376
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1377 This value is B<--CompoundIDMode> specific and indicates how compound ID is generated.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1378
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1379 For I<DataField> value of B<--CompoundIDMode> option, it corresponds to datafield label name
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1380 whose value is used as compound ID; otherwise, it's a prefix string used for generating compound
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1381 IDs like LabelPrefixString<Number>. Default value, I<Cmpd>, generates compound IDs which
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1382 look like Cmpd<Number>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1383
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1384 Examples for I<DataField> value of B<--CompoundIDMode>:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1385
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1386 MolID
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1387 ExtReg
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1388
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1389 Examples for I<LabelPrefix> or I<MolNameOrLabelPrefix> value of B<--CompoundIDMode>:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1390
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1391 Compound
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1392
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1393 The value specified above generates compound IDs which correspond to Compound<Number>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1394 instead of default value of Cmpd<Number>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1395
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1396 =item B<--CompoundIDLabel> I<text>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1397
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1398 Specify compound ID column label for CSV/TSV text file(s) used during I<CompoundID> value
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1399 of B<--DataFieldsMode> option. Default value: I<CompoundID>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1400
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1401 =item B<--CompoundIDMode> I<DataField | MolName | LabelPrefix | MolNameOrLabelPrefix>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1402
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1403 Specify how to generate compound IDs and write to CSV/TSV text file(s) along with calculated
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1404 physicochemical properties for I<text | both> values of B<--output> option: use a I<SDFile(s)>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1405 datafield value; use molname line from I<SDFile(s)>; generate a sequential ID with specific prefix;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1406 use combination of both MolName and LabelPrefix with usage of LabelPrefix values for empty
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1407 molname lines.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1408
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1409 Possible values: I<DataField | MolName | LabelPrefix | MolNameOrLabelPrefix>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1410 Default value: I<LabelPrefix>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1411
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1412 For I<MolNameAndLabelPrefix> value of B<--CompoundIDMode>, molname line in I<SDFile(s)> takes
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1413 precedence over sequential compound IDs generated using I<LabelPrefix> and only empty molname
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1414 values are replaced with sequential compound IDs.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1415
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1416 This is only used for I<CompoundID> value of B<--DataFieldsMode> option.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1417
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1418 =item B<--DataFields> I<"FieldLabel1,FieldLabel2,...">
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1419
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1420 Comma delimited list of I<SDFiles(s)> data fields to extract and write to CSV/TSV text file(s) along
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1421 with calculated physicochemical properties for I<text | both> values of B<--output> option.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1422
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1423 This is only used for I<Specify> value of B<--DataFieldsMode> option.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1424
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1425 Examples:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1426
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1427 Extreg
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1428 MolID,CompoundName
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1429
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1430 =item B<-d, --DataFieldsMode> I<All | Common | Specify | CompoundID>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1431
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1432 Specify how data fields in I<SDFile(s)> are transferred to output CSV/TSV text file(s) along
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1433 with calculated physicochemical properties for I<text | both> values of B<--output> option:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1434 transfer all SD data field; transfer SD data files common to all compounds; extract specified
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1435 data fields; generate a compound ID using molname line, a compound prefix, or a combination
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1436 of both. Possible values: I<All | Common | specify | CompoundID>. Default value: I<CompoundID>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1437
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1438 =item B<-f, --Filter> I<Yes | No>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1439
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1440 Specify whether to check and filter compound data in SDFile(s). Possible values: I<Yes or No>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1441 Default value: I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1442
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1443 By default, compound data is checked before calculating physiochemical properties and compounds
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1444 containing atom data corresponding to non-element symbols or no atom data are ignored.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1445
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1446 =item B<-h, --help>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1447
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1448 Print this help message.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1449
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1450 =item B<--HydrogenBonds> I<HBondsType1 | HBondsType2>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1451
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1452 Parameters to control calculation of hydrogen bond donors and acceptors. Possible values:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1453 I<HBondsType1, HydrogenBondsType1, HBondsType2, HydrogenBondsType2>. Default value:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1454 I<HBondsType2> which corresponds to B<RuleOf5> definition for number of hydrogen bond
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1455 donors and acceptors.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1456
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1457 The current release of MayaChemTools supports identification of two types of hydrogen bond
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1458 donor and acceptor atoms with these names:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1459
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1460 HBondsType1 or HydrogenBondsType1
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1461 HBondsType2 or HydrogenBondsType2
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1462
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1463 The names of these hydrogen bond types are rather arbitrary. However, their definitions have
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1464 specific meaning and are as follows:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1465
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1466 HydrogenBondsType1 [ Ref 60-61, Ref 65-66 ]:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1467
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1468 Donor: NH, NH2, OH - Any N and O with available H
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1469 Acceptor: N[!H], O - Any N without available H and any O
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1470
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1471 HydrogenBondsType2 [ Ref 91 ]:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1472
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1473 Donor: NH, NH2, OH - N and O with available H
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1474 Acceptor: N, O - And N and O
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1475
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1476 =item B<-k, --KeepLargestComponent> I<Yes | No>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1477
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1478 Calculate physicochemical properties for only the largest component in molecule. Possible values:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1479 I<Yes or No>. Default value: I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1480
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1481 For molecules containing multiple connected components, physicochemical properties can be
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1482 calculated in two different ways: use all connected components or just the largest connected
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1483 component. By default, all atoms except for the largest connected component are
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1484 deleted before calculation of physicochemical properties.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1485
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1486 =item B<-m, --mode> I<All | RuleOf5 | RuleOf3 | "name1, [name2,...]">
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1487
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1488 Specify physicochemical properties to calculate for SDFile(s): calculate all available physical
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1489 chemical properties; calculate properties corresponding to Rule of 5; or use a comma delimited
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1490 list of supported physicochemical properties. Possible values: I<All | RuleOf5 | RuleOf3 |
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1491 "name1, [name2,...]">.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1492
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1493 Default value: I<MolecularWeight, HeavyAtoms, MolecularVolume, RotatableBonds, HydrogenBondDonors,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1494 HydrogenBondAcceptors, SLogP, TPSA>. These properties are calculated by default.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1495
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1496 I<RuleOf5> [ Ref 91 ] includes these properties: I<MolecularWeight, HydrogenBondDonors, HydrogenBondAcceptors,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1497 SLogP>. I<RuleOf5> states: MolecularWeight <= 500, HydrogenBondDonors <= 5, HydrogenBondAcceptors <= 10, and
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1498 logP <= 5.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1499
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1500 I<RuleOf3> [ Ref 92 ] includes these properties: I<MolecularWeight, RotatableBonds, HydrogenBondDonors,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1501 HydrogenBondAcceptors, SLogP, TPSA>. I<RuleOf3> states: MolecularWeight <= 300, RotatableBonds <= 3,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1502 HydrogenBondDonors <= 3, HydrogenBondAcceptors <= 3, logP <= 3, and TPSA <= 60.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1503
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1504 I<All> calculates all supported physicochemical properties: I<MolecularWeight, ExactMass,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1505 HeavyAtoms, Rings, AromaticRings, MolecularVolume, RotatableBonds, HydrogenBondDonors,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1506 HydrogenBondAcceptors, SLogP, SMR, TPSA, Fsp3Carbons, Sp3Carbons, MolecularComplexity>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1507
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1508 =item B<--MolecularComplexity> I<Name,Value, [Name,Value,...]>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1509
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1510 Parameters to control calculation of molecular complexity: it's a comma delimited list of parameter
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1511 name and value pairs.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1512
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1513 Possible parameter names: I<MolecularComplexityType, AtomIdentifierType,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1514 AtomicInvariantsToUse, FunctionalClassesToUse, MACCSKeysSize, NeighborhoodRadius,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1515 MinPathLength, MaxPathLength, UseBondSymbols, MinDistance, MaxDistance,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1516 UseTriangleInequality, DistanceBinSize, NormalizationMethodology>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1517
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1518 The valid paramater valuse for each parameter name are described in the following sections.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1519
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1520 The current release of MayaChemTools supports calculation of molecular complexity using
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1521 I<MolecularComplexityType> parameter corresponding to the number of bits-set or unique
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1522 keys [ Ref 117-119 ] in molecular fingerprints. The valid values for I<MolecularComplexityType>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1523 are:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1524
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1525 AtomTypesFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1526 ExtendedConnectivityFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1527 MACCSKeys
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1528 PathLengthFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1529 TopologicalAtomPairsFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1530 TopologicalAtomTripletsFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1531 TopologicalAtomTorsionsFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1532 TopologicalPharmacophoreAtomPairsFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1533 TopologicalPharmacophoreAtomTripletsFingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1534
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1535 Default value for I<MolecularComplexityType>: I<MACCSKeys>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1536
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1537 I<AtomIdentifierType> parameter name correspods to atom types used during generation of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1538 fingerprints. The valid values for I<AtomIdentifierType> are: I<AtomicInvariantsAtomTypes,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1539 DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1540 SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes>. I<AtomicInvariantsAtomTypes>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1541 is not supported for during the following values of I<MolecularComplexityType>: I<MACCSKeys,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1542 TopologicalPharmacophoreAtomPairsFingerprints, TopologicalPharmacophoreAtomTripletsFingerprints>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1543 I<FunctionalClassAtomTypes> is the only valid value for I<AtomIdentifierType> for topological
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1544 pharmacophore fingerprints.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1545
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1546 Default value for I<AtomIdentifierType>: I<AtomicInvariantsAtomTypes>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1547 for all except topological pharmacophore fingerprints where it is I<FunctionalClassAtomTypes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1548
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1549 I<AtomicInvariantsToUse> parameter name and values are used during I<AtomicInvariantsAtomTypes>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1550 value of parameter I<AtomIdentifierType>. It's a list of space separated valid atomic invariant atom types.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1551
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1552 Possible values for atomic invariants are: I<AS, X, BO, LBO, SB, DB, TB, H, Ar, RA, FC, MN, SM>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1553 Default value for I<AtomicInvariantsToUse> parameter are set differently for different fingerprints
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1554 using I<MolecularComplexityType> parameter as shown below:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1555
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1556 MolecularComplexityType AtomicInvariantsToUse
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1557
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1558 AtomTypesFingerprints AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1559 TopologicalAtomPairsFingerprints AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1560 TopologicalAtomTripletsFingerprints AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1561 TopologicalAtomTorsionsFingerprints AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1562
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1563 ExtendedConnectivityFingerprints AS X BO H FC MN
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1564 PathLengthFingerprints AS
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1565
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1566
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1567 The atomic invariants abbreviations correspond to:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1568
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1569 AS = Atom symbol corresponding to element symbol
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1570
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1571 X<n> = Number of non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1572 BO<n> = Sum of bond orders to non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1573 LBO<n> = Largest bond order of non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1574 SB<n> = Number of single bonds to non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1575 DB<n> = Number of double bonds to non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1576 TB<n> = Number of triple bonds to non-hydrogen atom neighbors or heavy atoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1577 H<n> = Number of implicit and explicit hydrogens for atom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1578 Ar = Aromatic annotation indicating whether atom is aromatic
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1579 RA = Ring atom annotation indicating whether atom is a ring
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1580 FC<+n/-n> = Formal charge assigned to atom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1581 MN<n> = Mass number indicating isotope other than most abundant isotope
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1582 SM<n> = Spin multiplicity of atom. Possible values: 1 (singlet), 2 (doublet) or
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1583 3 (triplet)
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1584
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1585 Atom type generated by AtomTypes::AtomicInvariantsAtomTypes class corresponds to:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1586
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1587 AS.X<n>.BO<n>.LBO<n>.<SB><n>.<DB><n>.<TB><n>.H<n>.Ar.RA.FC<+n/-n>.MN<n>.SM<n>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1588
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1589 Except for AS which is a required atomic invariant in atom types, all other atomic invariants are
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1590 optional. Atom type specification doesn't include atomic invariants with zero or undefined values.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1591
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1592 In addition to usage of abbreviations for specifying atomic invariants, the following descriptive words
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1593 are also allowed:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1594
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1595 X : NumOfNonHydrogenAtomNeighbors or NumOfHeavyAtomNeighbors
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1596 BO : SumOfBondOrdersToNonHydrogenAtoms or SumOfBondOrdersToHeavyAtoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1597 LBO : LargestBondOrderToNonHydrogenAtoms or LargestBondOrderToHeavyAtoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1598 SB : NumOfSingleBondsToNonHydrogenAtoms or NumOfSingleBondsToHeavyAtoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1599 DB : NumOfDoubleBondsToNonHydrogenAtoms or NumOfDoubleBondsToHeavyAtoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1600 TB : NumOfTripleBondsToNonHydrogenAtoms or NumOfTripleBondsToHeavyAtoms
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1601 H : NumOfImplicitAndExplicitHydrogens
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1602 Ar : Aromatic
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1603 RA : RingAtom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1604 FC : FormalCharge
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1605 MN : MassNumber
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1606 SM : SpinMultiplicity
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1607
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1608 I<AtomTypes::AtomicInvariantsAtomTypes> module is used to assign atomic invariant
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1609 atom types.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1610
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1611 I<FunctionalClassesToUse> parameter name and values are used during I<FunctionalClassAtomTypes>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1612 value of parameter I<AtomIdentifierType>. It's a list of space separated valid atomic invariant atom types.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1613
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1614 Possible values for atom functional classes are: I<Ar, CA, H, HBA, HBD, Hal, NI, PI, RA>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1615
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1616 Default value for I<FunctionalClassesToUse> parameter is set to:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1617
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1618 HBD HBA PI NI Ar Hal
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1619
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1620 for all fingerprints except for the following two I<MolecularComplexityType> fingerints:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1621
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1622 MolecularComplexityType FunctionalClassesToUse
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1623
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1624 TopologicalPharmacophoreAtomPairsFingerprints HBD HBA P, NI H
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1625 TopologicalPharmacophoreAtomTripletsFingerprints HBD HBA PI NI H Ar
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1626
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1627 The functional class abbreviations correspond to:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1628
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1629 HBD: HydrogenBondDonor
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1630 HBA: HydrogenBondAcceptor
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1631 PI : PositivelyIonizable
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1632 NI : NegativelyIonizable
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1633 Ar : Aromatic
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1634 Hal : Halogen
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1635 H : Hydrophobic
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1636 RA : RingAtom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1637 CA : ChainAtom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1638
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1639 Functional class atom type specification for an atom corresponds to:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1640
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1641 Ar.CA.H.HBA.HBD.Hal.NI.PI.RA
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1642
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1643 I<AtomTypes::FunctionalClassAtomTypes> module is used to assign functional class atom
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1644 types. It uses following definitions [ Ref 60-61, Ref 65-66 ]:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1645
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1646 HydrogenBondDonor: NH, NH2, OH
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1647 HydrogenBondAcceptor: N[!H], O
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1648 PositivelyIonizable: +, NH2
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1649 NegativelyIonizable: -, C(=O)OH, S(=O)OH, P(=O)OH
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1650
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1651 I<MACCSKeysSize> parameter name is only used during I<MACCSKeys> value of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1652 I<MolecularComplexityType> and corresponds to the size of MACCS key set. Possible
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1653 values: I<166 or 322>. Default value: I<166>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1654
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1655 I<NeighborhoodRadius> parameter name is only used during I<ExtendedConnectivityFingerprints>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1656 value of I<MolecularComplexityType> and corresponds to atomic neighborhoods radius for
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1657 generating extended connectivity fingerprints. Possible values: positive integer. Default value:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1658 I<2>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1659
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1660 I<MinPathLength> and I<MaxPathLength> parameters are only used during I<PathLengthFingerprints>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1661 value of I<MolecularComplexityType> and correspond to minimum and maximum path lengths to use
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1662 for generating path length fingerprints. Possible values: positive integers. Default value: I<MinPathLength - 1>;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1663 I<MaxPathLength - 8>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1664
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1665 I<UseBondSymbols> parameter is only used during I<PathLengthFingerprints> value of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1666 I<MolecularComplexityType> and indicates whether bond symbols are included in atom path
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1667 strings used to generate path length fingerprints. Possible value: I<Yes or No>. Default value:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1668 I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1669
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1670 I<MinDistance> and I<MaxDistance> parameters are only used during I<TopologicalAtomPairsFingerprints>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1671 and I<TopologicalAtomTripletsFingerprints> values of I<MolecularComplexityType> and correspond to
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1672 minimum and maximum bond distance between atom pairs during topological pharmacophore fingerprints.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1673 Possible values: positive integers. Default value: I<MinDistance - 1>; I<MaxDistance - 10>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1674
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1675 I<UseTriangleInequality> parameter is used during these values for I<MolecularComplexityType>:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1676 I<TopologicalAtomTripletsFingerprints> and I<TopologicalPharmacophoreAtomTripletsFingerprints>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1677 Possible values: I<Yes or No>. It determines wheter to apply triangle inequality to distance triplets.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1678 Default value: I<TopologicalAtomTripletsFingerprints - No>;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1679 I<TopologicalPharmacophoreAtomTripletsFingerprints - Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1680
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1681 I<DistanceBinSize> parameter is used during I<TopologicalPharmacophoreAtomTripletsFingerprints>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1682 value of I<MolecularComplexityType> and correspons to distance bin size used for binning
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1683 distances during generation of topological pharmacophore atom triplets fingerprints. Possible
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1684 value: positive integer. Default value: I<2>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1685
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1686 I<NormalizationMethodology> is only used for these values for I<MolecularComplexityType>:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1687 I<ExtendedConnectivityFingerprints>, I<TopologicalPharmacophoreAtomPairsFingerprints>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1688 and I<TopologicalPharmacophoreAtomTripletsFingerprints>. It corresponds to normalization
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1689 methodology to use for scaling the number of bits-set or unique keys during generation of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1690 fingerprints. Possible values during I<ExtendedConnectivityFingerprints>: I<None or
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1691 ByHeavyAtomsCount>; Default value: I<None>. Possible values during topological
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1692 pharmacophore atom pairs and tripletes fingerprints: I<None or ByPossibleKeysCount>;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1693 Default value: I<None>. I<ByPossibleKeysCount> corresponds to total number of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1694 possible topological pharmacophore atom pairs or triplets in a molecule.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1695
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1696 Examples of I<MolecularComplexity> name and value parameters:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1697
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1698 MolecularComplexityType,AtomTypesFingerprints,AtomIdentifierType,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1699 AtomicInvariantsAtomTypes,AtomicInvariantsToUse,AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1700
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1701 MolecularComplexityType,ExtendedConnectivityFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1702 AtomIdentifierType,AtomicInvariantsAtomTypes,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1703 AtomicInvariantsToUse,AS X BO H FC MN,NeighborhoodRadius,2,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1704 NormalizationMethodology,None
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1705
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1706 MolecularComplexityType,MACCSKeys,MACCSKeysSize,166
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1707
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1708 MolecularComplexityType,PathLengthFingerprints,AtomIdentifierType,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1709 AtomicInvariantsAtomTypes,AtomicInvariantsToUse,AS,MinPathLength,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1710 1,MaxPathLength,8,UseBondSymbols,Yes
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1711
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1712 MolecularComplexityType,TopologicalAtomPairsFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1713 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1714 AS X BO H FC,MinDistance,1,MaxDistance,10
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1715
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1716 MolecularComplexityType,TopologicalAtomTripletsFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1717 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1718 AS X BO H FC,MinDistance,1,MaxDistance,10,UseTriangleInequality,No
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1719
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1720 MolecularComplexityType,TopologicalAtomTorsionsFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1721 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1722 AS X BO H FC
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1723
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1724 MolecularComplexityType,TopologicalPharmacophoreAtomPairsFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1725 AtomIdentifierType,FunctionalClassAtomTypes,FunctionalClassesToUse,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1726 HBD HBA PI NI H,MinDistance,1,MaxDistance,10,NormalizationMethodology,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1727 None
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1728
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1729 MolecularComplexityType,TopologicalPharmacophoreAtomTripletsFingerprints,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1730 AtomIdentifierType,FunctionalClassAtomTypes,FunctionalClassesToUse,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1731 HBD HBA PI NI H Ar,MinDistance,1,MaxDistance,10,NormalizationMethodology,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1732 None,UseTriangleInequality,Yes,NormalizationMethodology,None,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1733 DistanceBinSize,2
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1734
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1735 =item B<--OutDelim> I<comma | tab | semicolon>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1736
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1737 Delimiter for output CSV/TSV text file(s). Possible values: I<comma, tab, or semicolon>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1738 Default value: I<comma>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1739
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1740 =item B<--output> I<SD | text | both>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1741
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1742 Type of output files to generate. Possible values: I<SD, text, or both>. Default value: I<text>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1743
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1744 =item B<-o, --overwrite>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1745
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1746 Overwrite existing files.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1747
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1748 =item B<--Precision> I<Name,Number,[Name,Number,..]>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1749
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1750 Precision of calculated property values in the output file: it's a comma delimited list of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1751 property name and precision value pairs. Possible property names: I<MolecularWeight,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1752 ExactMass>. Possible values: positive intergers. Default value: I<MolecularWeight,2,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1753 ExactMass,4>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1754
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1755 Examples:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1756
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1757 ExactMass,3
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1758 MolecularWeight,1,ExactMass,2
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1759
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1760 =item B<-q, --quote> I<Yes | No>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1761
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1762 Put quote around column values in output CSV/TSV text file(s). Possible values:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1763 I<Yes or No>. Default value: I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1764
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1765 =item B<-r, --root> I<RootName>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1766
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1767 New file name is generated using the root: <Root>.<Ext>. Default for new file names:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1768 <SDFileName><PhysicochemicalProperties>.<Ext>. The file type determines <Ext> value.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1769 The sdf, csv, and tsv <Ext> values are used for SD, comma/semicolon, and tab
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1770 delimited text files, respectively.This option is ignored for multiple input files.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1771
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1772 =item B<--RotatableBonds> I<Name,Value, [Name,Value,...]>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1773
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1774 Parameters to control calculation of rotatable bonds [ Ref 92 ]: it's a comma delimited list of parameter
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1775 name and value pairs. Possible parameter names: I<IgnoreTerminalBonds, IgnoreBondsToTripleBonds,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1776 IgnoreAmideBonds, IgnoreThioamideBonds, IgnoreSulfonamideBonds>. Possible parameter values:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1777 I<Yes or No>. By default, value of all parameters is set to I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1778
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1779 =item B<--RuleOf3Violations> I<Yes | No>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1780
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1781 Specify whether to calculate B<RuleOf3Violations> for SDFile(s). Possible values: I<Yes or No>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1782 Default value: I<No>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1783
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1784 For I<Yes> value of B<RuleOf3Violations>, in addition to calculating total number of B<RuleOf3> violations,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1785 individual violations for compounds are also written to output files.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1786
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1787 B<RuleOf3> [ Ref 92 ] states: MolecularWeight <= 300, RotatableBonds <= 3, HydrogenBondDonors <= 3,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1788 HydrogenBondAcceptors <= 3, logP <= 3, and TPSA <= 60.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1789
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1790 =item B<--RuleOf5Violations> I<Yes | No>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1791
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1792 Specify whether to calculate B<RuleOf5Violations> for SDFile(s). Possible values: I<Yes or No>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1793 Default value: I<No>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1794
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1795 For I<Yes> value of B<RuleOf5Violations>, in addition to calculating total number of B<RuleOf5> violations,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1796 individual violations for compounds are also written to output files.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1797
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1798 B<RuleOf5> [ Ref 91 ] states: MolecularWeight <= 500, HydrogenBondDonors <= 5, HydrogenBondAcceptors <= 10,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1799 and logP <= 5.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1800
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1801 =item B<--TPSA> I<Name,Value, [Name,Value,...]>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1802
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1803 Parameters to control calculation of TPSA: it's a comma delimited list of parameter name and value
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1804 pairs. Possible parameter names: I<IgnorePhosphorus, IgnoreSulfur>. Possible parameter values:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1805 I<Yes or No>. By default, value of all parameters is set to I<Yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1806
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1807 By default, TPSA atom contributions from Phosphorus and Sulfur atoms are not included during
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1808 TPSA calculations. [ Ref 91 ]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1809
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1810 =item B<-w, --WorkingDir> I<DirName>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1811
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1812 Location of working directory. Default value: current directory.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1813
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1814 =back
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1815
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1816 =head1 EXAMPLES
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1817
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1818 To calculate default set of physicochemical properties - MolecularWeight, HeavyAtoms,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1819 MolecularVolume, RotatableBonds, HydrogenBondDonor, HydrogenBondAcceptors, SLogP,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1820 TPSA - and generate a SamplePhysicochemicalProperties.csv file containing sequential
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1821 compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1822
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1823 % CalculatePhysicochemicalProperties.pl -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1824
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1825 To calculate all available physicochemical properties and generate both SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1826 and SampleAllProperties.sdf files containing sequential compound IDs in CSV file along with
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1827 properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1828
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1829 % CalculatePhysicochemicalProperties.pl -m All --output both
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1830 -r SampleAllProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1831
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1832 To calculate RuleOf5 physicochemical properties and generate a SampleRuleOf5Properties.csv file
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1833 containing sequential compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1834
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1835 % CalculatePhysicochemicalProperties.pl -m RuleOf5
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1836 -r SampleRuleOf5Properties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1837
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1838 To calculate RuleOf5 physicochemical properties along with counting RuleOf5 violations and generate
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1839 a SampleRuleOf5Properties.csv file containing sequential compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1840
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1841 % CalculatePhysicochemicalProperties.pl -m RuleOf5 --RuleOf5Violations Yes
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1842 -r SampleRuleOf5Properties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1843
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1844 To calculate RuleOf3 physicochemical properties and generate a SampleRuleOf3Properties.csv file
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1845 containing sequential compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1846
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1847 % CalculatePhysicochemicalProperties.pl -m RuleOf3
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1848 -r SampleRuleOf3Properties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1849
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1850 To calculate RuleOf3 physicochemical properties along with counting RuleOf3 violations and generate
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1851 a SampleRuleOf3Properties.csv file containing sequential compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1852
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1853 % CalculatePhysicochemicalProperties.pl -m RuleOf3 --RuleOf3Violations Yes
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1854 -r SampleRuleOf3Properties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1855
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1856 To calculate a specific set of physicochemical properties and generate a SampleProperties.csv file
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1857 containing sequential compound IDs along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1858
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1859 % CalculatePhysicochemicalProperties.pl -m "Rings,AromaticRings"
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1860 -r SampleProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1861
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1862 To calculate HydrogenBondDonors and HydrogenBondAcceptors using HydrogenBondsType1 definition
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1863 and generate a SampleProperties.csv file containing sequential compound IDs along with properties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1864 data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1865
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1866 % CalculatePhysicochemicalProperties.pl -m "HydrogenBondDonors,HydrogenBondAcceptors"
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1867 --HydrogenBonds HBondsType1 -r SampleProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1868
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1869 To calculate TPSA using sulfur and phosphorus atoms along with nitrogen and oxygen atoms and
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1870 generate a SampleProperties.csv file containing sequential compound IDs along with properties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1871 data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1872
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1873 % CalculatePhysicochemicalProperties.pl -m "TPSA" --TPSA "IgnorePhosphorus,No,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1874 IgnoreSulfur,No" -r SampleProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1875
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1876 To calculate MolecularComplexity using extendend connectivity fingerprints corresponding
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1877 to atom neighborhood radius of 2 with atomic invariant atom types without any scaling and
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1878 generate a SampleProperties.csv file containing sequential compound IDs along with properties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1879 data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1880
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1881 % CalculatePhysicochemicalProperties.pl -m MolecularComplexity --MolecularComplexity
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1882 "MolecularComplexityType,ExtendedConnectivityFingerprints,NeighborhoodRadius,2,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1883 AtomIdentifierType, AtomicInvariantsAtomTypes,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1884 AtomicInvariantsToUse,AS X BO H FC MN,NormalizationMethodology,None"
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1885 -r SampleProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1886
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1887 To calculate RuleOf5 physicochemical properties along with counting RuleOf5 violations and generate
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1888 a SampleRuleOf5Properties.csv file containing compound IDs from molecule name line along with
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1889 properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1890
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1891 % CalculatePhysicochemicalProperties.pl -m RuleOf5 --RuleOf5Violations Yes
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1892 --DataFieldsMode CompoundID --CompoundIDMode MolName
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1893 -r SampleRuleOf5Properties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1894
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1895 To calculate all available physicochemical properties and generate a SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1896 file containing compound ID using specified data field along with along with properties data,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1897 type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1898
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1899 % CalculatePhysicochemicalProperties.pl -m All
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1900 --DataFieldsMode CompoundID --CompoundIDMode DataField --CompoundID Mol_ID
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1901 -r SampleAllProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1902
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1903 To calculate all available physicochemical properties and generate a SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1904 file containing compound ID using combination of molecule name line and an explicit compound
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1905 prefix along with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1906
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1907 % CalculatePhysicochemicalProperties.pl -m All
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1908 --DataFieldsMode CompoundID --CompoundIDMode MolnameOrLabelPrefix
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1909 --CompoundID Cmpd --CompoundIDLabel MolID -r SampleAllProperties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1910 -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1911
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1912 To calculate all available physicochemical properties and generate a SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1913 file containing specific data fields columns along with with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1914
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1915 % CalculatePhysicochemicalProperties.pl -m All
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1916 --DataFieldsMode Specify --DataFields Mol_ID -r SampleAllProperties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1917 -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1918
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1919 To calculate all available physicochemical properties and generate a SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1920 file containing common data fields columns along with with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1921
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1922 % CalculatePhysicochemicalProperties.pl -m All
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1923 --DataFieldsMode Common -r SampleAllProperties -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1924
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1925 To calculate all available physicochemical properties and generate both SampleAllProperties.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1926 and CSV files containing all data fields columns in CSV files along with with properties data, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1927
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1928 % CalculatePhysicochemicalProperties.pl -m All
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1929 --DataFieldsMode All --output both -r SampleAllProperties
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1930 -o Sample.sdf
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1931
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1932 =head1 AUTHOR
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1933
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1934 Manish Sud <msud@san.rr.com>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1935
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1936 =head1 SEE ALSO
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1937
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1938 ExtractFromSDtFiles.pl, ExtractFromTextFiles.pl, InfoSDFiles.pl, InfoTextFiles.pl
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1939
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1940 =head1 COPYRIGHT
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1941
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1942 Copyright (C) 2015 Manish Sud. All rights reserved.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1943
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1944 This file is part of MayaChemTools.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1945
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1946 MayaChemTools is free software; you can redistribute it and/or modify it under
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1947 the terms of the GNU Lesser General Public License as published by the Free
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1948 Software Foundation; either version 3 of the License, or (at your option)
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1949 any later version.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1950
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1951 =cut