annotate bin/SDToMolFiles.pl @ 0:4816e4a8ae95 draft default tip

Uploaded
author deepakjadmin
date Wed, 20 Jan 2016 09:23:18 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
1 #!/usr/bin/perl -w
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
2 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
3 # $RCSfile: SDToMolFiles.pl,v $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
4 # $Date: 2015/02/28 20:46:20 $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
5 # $Revision: 1.35 $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
6 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
7 # Author: Manish Sud <msud@san.rr.com>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
8 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
9 # Copyright (C) 2015 Manish Sud. All rights reserved.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
10 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
11 # This file is part of MayaChemTools.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
12 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
13 # MayaChemTools is free software; you can redistribute it and/or modify it under
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
14 # the terms of the GNU Lesser General Public License as published by the Free
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
15 # Software Foundation; either version 3 of the License, or (at your option) any
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
16 # later version.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
17 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
18 # MayaChemTools is distributed in the hope that it will be useful, but without
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
19 # any warranty; without even the implied warranty of merchantability of fitness
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
20 # for a particular purpose. See the GNU Lesser General Public License for more
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
21 # details.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
22 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
23 # You should have received a copy of the GNU Lesser General Public License
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
26 # Boston, MA, 02111-1307, USA.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
27 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
28
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
29 use strict;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
30 use FindBin; use lib "$FindBin::Bin/../lib";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
31 use Getopt::Long;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
32 use File::Basename;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
33 use Text::ParseWords;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
34 use Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
35 use SDFileUtil;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
36 use FileUtil;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
37
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
38 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
39
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
40 # Autoflush STDOUT
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
41 $| = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
42
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
43 # Starting message...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
44 $ScriptName = basename $0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
45 print "\n$ScriptName:Starting...\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
46 $StartTime = new Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
47
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
48 # Get the options and setup script...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
49 SetupScriptUsage();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
50 if ($Options{help} || @ARGV < 1) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
51 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
52 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
53
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
54 my(@SDFilesList);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
55 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd");
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
56
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
57 # Process options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
58 print "Processing options...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
59 my(%OptionsInfo);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
60 ProcessOptions();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
61
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
62 # Setup information about input files...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
63 print "Checking input SD file(s)...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
64 my(%SDFilesInfo);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
65 RetrieveSDFilesInfo();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
66
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
67 # Process input files..
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
68 my($FileIndex);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
69 if (@SDFilesList > 1) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
70 print "\nProcessing SD files...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
71 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
72 for $FileIndex (0 .. $#SDFilesList) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
73 if ($SDFilesInfo{FileOkay}[$FileIndex]) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
74 print "\nProcessing file $SDFilesList[$FileIndex]...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
75 GenerateMolFiles($FileIndex);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
76 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
77 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
78 print "\n$ScriptName:Done...\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
79
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
80 $EndTime = new Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
81 $TotalTime = timediff ($EndTime, $StartTime);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
82 print "Total time: ", timestr($TotalTime), "\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
83
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
84 ###############################################################################
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
85
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
86 # Generate MOL files for a SD file...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
87 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
88 sub GenerateMolFiles {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
89 my($FileIndex) = @_;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
90 my($SDFile, $MOLFile, $MOLFileRoot, $OutFileRoot, $OverwriteFiles, $UseDataField, $DataFieldName, $UseMolName, $CmpdCount, $MolEndDelimiter, $CmpdString, @CmpdLines, %DataFieldValues);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
91
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
92 $SDFile = $SDFilesList[$FileIndex];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
93
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
94 if (!open SDFILE, "$SDFile") {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
95 warn "Warning: Ignoring file $SDFile: Couldn't open it: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
96 return;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
97 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
98
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
99 $CmpdCount = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
100 $MolEndDelimiter = "M END";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
101
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
102 $OutFileRoot = $SDFilesInfo{OutFileRoot}[$FileIndex];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
103 $OverwriteFiles = $OptionsInfo{OverwriteFiles};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
104
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
105 $UseDataField = ($OptionsInfo{Mode} =~ /^DataField$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
106 $DataFieldName = $OptionsInfo{DataField};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
107
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
108 $UseMolName = ($OptionsInfo{Mode} =~ /^MolName$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
109
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
110 CMPDSTRING: while ($CmpdString = ReadCmpdString(\*SDFILE)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
111 $CmpdCount++;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
112
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
113 # Setup MOL file name...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
114 $MOLFileRoot = '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
115 if ($UseDataField) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
116 @CmpdLines = split "\n", $CmpdString;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
117 %DataFieldValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
118 if (exists $DataFieldValues{$DataFieldName}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
119 $MOLFileRoot = $DataFieldValues{$DataFieldName};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
120 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
121 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
122 elsif ($UseMolName) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
123 @CmpdLines = split "\n", $CmpdString;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
124 $MOLFileRoot = $CmpdLines[0];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
125 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
126
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
127 # Check for any invalid file name characters in data field or molname values...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
128 if ($MOLFileRoot && $MOLFileRoot =~ /[^a-zA-Z0-9_]/) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
129 $MOLFileRoot =~ s/[^a-zA-Z0-9_]//g;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
130 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
131 # Fall back plan for MOL file name...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
132 if (!$MOLFileRoot) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
133 $MOLFileRoot = "${OutFileRoot}Cmpd${CmpdCount}";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
134 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
135
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
136 $MOLFile = "${MOLFileRoot}.mol";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
137
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
138 if (!$OverwriteFiles) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
139 if (-e $MOLFile) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
140 warn "Warning: Ignoring compound number, $CmpdCount, in $SDFile: New MOL file, $MOLFile, already exists\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
141 next CMPDSTRING;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
142 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
143 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
144
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
145 if (!($CmpdString =~ /$MolEndDelimiter/)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
146 warn "Warning: Ignoring compound number, $CmpdCount, in $SDFile: Invalid compound data\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
147 next CMPDSTRING;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
148 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
149
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
150 # Write out MOL file...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
151
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
152 print "Generating $MOLFile file...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
153 open MOLFILE, ">$MOLFile" or die "Error: Can't open $MOLFile: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
154 ($CmpdString) = split "$MolEndDelimiter", $CmpdString;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
155 print MOLFILE "$CmpdString";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
156 print MOLFILE "$MolEndDelimiter\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
157 close MOLFILE;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
158
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
159 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
160
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
161 close SDFILE;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
162 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
163
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
164 # Retrieve information about SD files...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
165 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
166 sub RetrieveSDFilesInfo {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
167 my($SDFile, $Index, $FileDir, $FileName, $FileExt, $OutFileRoot);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
168
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
169 %SDFilesInfo = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
170 @{$SDFilesInfo{FileOkay}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
171 @{$SDFilesInfo{OutFileRoot}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
172
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
173 FILELIST: for $Index (0 .. $#SDFilesList) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
174 $SDFile = $SDFilesList[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
175
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
176 $SDFilesInfo{FileOkay}[$Index] = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
177 $SDFilesInfo{OutFileRoot}[$Index] = '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
178
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
179 $SDFile = $SDFilesList[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
180 if (!(-e $SDFile)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
181 warn "Warning: Ignoring file $SDFile: It doesn't exist\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
182 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
183 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
184 if (!CheckFileType($SDFile, "sd sdf")) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
185 warn "Warning: Ignoring file $SDFile: It's not a SD file\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
186 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
187 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
188
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
189 # Setup output file root...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
190 $FileDir = ""; $FileName = ""; $FileExt = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
191 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
192
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
193 if ($OptionsInfo{OutFileRoot} && (@SDFilesList == 1)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
194 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($OptionsInfo{OutFileRoot});
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
195 if ($RootFileName && $RootFileExt) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
196 $FileName = $RootFileName;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
197 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
198 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
199 $FileName = $OptionsInfo{OutFileRoot};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
200 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
201 $OutFileRoot = $FileName;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
202 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
203 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
204 $OutFileRoot = "$FileName";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
205 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
206
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
207 $SDFilesInfo{FileOkay}[$Index] = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
208 $SDFilesInfo{OutFileRoot}[$Index] = $OutFileRoot;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
209 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
210 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
211
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
212 # Process option values...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
213 sub ProcessOptions {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
214 %OptionsInfo = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
215
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
216 $OptionsInfo{Mode} = $Options{mode};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
217
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
218 $OptionsInfo{DataField} = '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
219 if ($Options{mode} =~ /^DataField$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
220 if (!$Options{datafield}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
221 die "Error: You must specify a value for \"-d, --DataField\" option in \"DataField\" \"-m, --mode\". \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
222 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
223 $OptionsInfo{DataField} = $Options{datafield};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
224 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
225
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
226 $OptionsInfo{OverwriteFiles} = $Options{overwrite} ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
227
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
228 $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
229 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
230
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
231 # Setup script usage and retrieve command line arguments specified using various options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
232 sub SetupScriptUsage {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
233
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
234 # Retrieve all the options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
235 %Options = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
236
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
237 $Options{mode} = 'RootPrefix';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
238
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
239 if (!GetOptions(\%Options, "datafield|d=s", "help|h", "mode|m=s", "overwrite|o", "root|r=s", "workingdir|w=s")) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
240 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
241 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
242 if ($Options{workingdir}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
243 if (! -d $Options{workingdir}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
244 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
245 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
246 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
247 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
248
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
249 if ($Options{mode} !~ /^(DataField|MolName|RootPrefix)$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
250 die "Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: DataField, MolName, RootPrefix\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
251 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
252 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
253
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
254 __END__
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
255
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
256 =head1 NAME
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
257
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
258 SDToMolFiles.pl - Generate MDLMOL file(s) from SD file(s)
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
259
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
260 =head1 SYNOPSIS
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
261
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
262 SDToMolFiles.pl SDFile(s)...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
263
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
264 SDToMolFiles.pl [B<-d, --DataField> DataFieldName]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
265 [B<-m, --mode> DataField | MolName | RootPrefix] [B<-h, --help>]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
266 [B<-o, --overwrite>] [B<-r, --root> rootname]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
267 [B<-w, --workingdir> dirname] SDFile(s)...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
268
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
269 =head1 DESCRIPTION
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
270
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
271 Generate MDLMOL file(s) from I<SDFile(s)>. All header data labels and values in
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
272 SDFile(s) are simply ignored; other appopriate data from SDFile(s) is transferred to MDLMOL
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
273 files. Multiple I<SDFile(s)> names are separated by spaces. The valid file extensions are
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
274 I<.sdf> and I<.sd>. All other file names are ignored. All the SD files in a current
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
275 directory can be specified either by I<*.sdf> or the current directory name.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
276
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
277 =head1 OPTIONS
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
278
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
279 =over 4
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
280
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
281 =item B<-d, --DataField> I<DataFieldName>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
282
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
283 Specify I<SDFile(s)> datafield label name whose value is used for generation of MDLMOL
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
284 file names. Default value: I<None>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
285
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
286 =item B<-h, --help>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
287
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
288 Print this help message.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
289
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
290 =item B<-m, --mode> I<DataField | MolName | RootPrefix>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
291
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
292 Specify how to generate MDLMOL file names: use a I<SDFile(s)> datafield value; use
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
293 molname line from I<SDFile(s)>; generate a sequential ID using root prefix specified
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
294 by B<-r, --root> option.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
295
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
296 Possible values: I<DataField | MolName | RootPrefix | RootPrefix>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
297 Default: I<RootPrefix>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
298
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
299 For empty I<MolName> and I<DataField> values during these specified modes, file
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
300 name is automatically generated using I<RootPrefix>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
301
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
302 For I<RootPrefix> value of B<-m, --mode> option, MDLMOL file names are generated
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
303 using by appending compound record number to value of B<-r, --root> option. For
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
304 example: I<RootName>Cmd<RecordNumber>.mol.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
305
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
306 Allowed characters in file names are: a-zA-Z0-9_. All other characters in datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
307 values, molname line, and root prefix are ignore during generation of file names.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
308
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
309 =item B<-o, --overwrite>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
310
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
311 Overwrite existing files.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
312
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
313 =item B<-r, --root> I<rootname>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
314
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
315 Specify root name to used during I<RootPrefix> B<-m, --mode> option value.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
316 New MDLMOL file names are generated using the root: <Root>Cmpd<RecordNumber>.mol
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
317 Default for new file names: <InitialSDFileName>Cmpd<RecordNumber>.mol. This option
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
318 is ignored for multiple input files.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
319
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
320 =item B<-w, --workingdir> I<dirname>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
321
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
322 Location of working directory. Default: current directory.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
323
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
324 =back
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
325
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
326 =head1 EXAMPLES
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
327
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
328 To generate MDLMOL files from Sample1*.sdf and Sample2*.sd files, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
329
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
330 % SDToMolFiles.pl -o Sample1*.sdf Sample2*.sd
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
331
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
332 To generate Sample*.mol files from Sample1.sdf, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
333
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
334 % SDToMolFiles.pl -r Sample -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
335
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
336 To generate MOL files from Sample1.sdf using molname line data for generating
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
337 MOL file names, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
338
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
339 % SDToMolFiles.pl -m MolName -r Sample -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
340
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
341 To generate MOL files from Sample1.sdf using a specific data field values for
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
342 generating MOL file names, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
343
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
344 % SDToMolFiles.pl -m DataField --DataField MolID -r Sample
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
345 -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
346
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
347 =head1 AUTHOR
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
348
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
349
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
350 =head1 AUTHOR
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
351
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
352 Manish Sud <msud@san.rr.com>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
353
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
354 =head1 SEE ALSO
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
355
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
356 InfoSDFiles.pl, MolFilesToSD.pl
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
357
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
358 =head1 COPYRIGHT
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
359
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
360 Copyright (C) 2015 Manish Sud. All rights reserved.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
361
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
362 This file is part of MayaChemTools.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
363
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
364 MayaChemTools is free software; you can redistribute it and/or modify it under
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
365 the terms of the GNU Lesser General Public License as published by the Free
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
366 Software Foundation; either version 3 of the License, or (at your option)
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
367 any later version.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
368
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
369 =cut