1 #!/usr/bin/perl -w 2 # 3 # $RCSfile: FilterSDFiles.pl,v $ 4 # $Date: 2008/01/30 21:44:46 $ 5 # $Revision: 1.19 $ 6 # 7 # Author: Manish Sud <msud@san.rr.com> 8 # 9 # Copyright (C) 2004-2008 Manish Sud. All rights reserved. 10 # 11 # This file is part of MayaChemTools. 12 # 13 # MayaChemTools is free software; you can redistribute it and/or modify it under 14 # the terms of the GNU Lesser General Public License as published by the Free 15 # Software Foundation; either version 3 of the License, or (at your option) any 16 # later version. 17 # 18 # MayaChemTools is distributed in the hope that it will be useful, but without 19 # any warranty; without even the implied warranty of merchantability of fitness 20 # for a particular purpose. See the GNU Lesser General Public License for more 21 # details. 22 # 23 # You should have received a copy of the GNU Lesser General Public License 24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or 25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330, 26 # Boston, MA, 02111-1307, USA. 27 # 28 29 use 5.006; 30 use strict; 31 use FindBin; use lib "$FindBin::Bin/../lib"; 32 use Getopt::Long; 33 use File::Basename; 34 use Benchmark; 35 use SDFileUtil; 36 use FileUtil; 37 38 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime); 39 my($SDFile, @SDFilesList, $NewSDFile, $NewKeepSDFile, $NewFileName, $Index, $FileDir, $FileName, $FileExt, $CmpdCount, $FilteredCmpdCount, $KeepCmpdCount, $CtabLinesCount, @CmpdLines, $CmpdString, $FilterCmpd, $PrintCmpdCounterHeader); 40 41 # Autoflush STDOUT 42 $| = 1; 43 44 # Starting message... 45 $ScriptName = basename $0; 46 print "\n$ScriptName:Starting...\n\n"; 47 $StartTime = new Benchmark; 48 49 # Get the options and setup script... 50 SetupScriptUsage(); 51 if ($Options{help} || @ARGV < 1) { 52 die GetUsageFromPod("$FindBin::Bin/$ScriptName"); 53 } 54 55 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd"); 56 57 # Process all the SD files... 58 if (@SDFilesList > 1) { 59 print "Processing SD files...\n"; 60 } 61 62 FILELIST: for $Index (0 .. $#SDFilesList) { 63 $SDFile = $SDFilesList[$Index]; 64 if (@SDFilesList > 1) { 65 print "\nProcessing file $SDFile...\n"; 66 } 67 else { 68 print "Processing file $SDFile...\n" 69 } 70 if (!(-e $SDFile)) { 71 warn "Warning: Ignoring file $SDFile: It doesn't exist\n"; 72 next FILELIST; 73 } 74 if (!CheckFileType($SDFile, "sd sdf")) { 75 warn "Warning: Ignoring file $SDFile: It's not a SD file\n"; 76 next FILELIST; 77 } 78 if (!open SDFILE, "$SDFile") { 79 warn "Warning: Ignoring file $SDFile: Couldn't open it: $! \n"; 80 next FILELIST; 81 } 82 83 # Setup new file name... 84 $FileDir = ""; $FileName = ""; $FileExt = ""; 85 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile); 86 if ($Options{root} && (@SDFilesList == 1)) { 87 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root}); 88 if ($RootFileName && $RootFileExt) { 89 $NewFileName = $RootFileName; 90 } 91 else { 92 $NewSDFile = $Options{root}; 93 } 94 $NewKeepSDFile = $NewSDFile; 95 } 96 else { 97 $NewSDFile = $FileName . "Filtered"; 98 $NewKeepSDFile = $FileName; 99 } 100 $NewSDFile .= ".$FileExt"; 101 $NewKeepSDFile .= "Ignored" . ".$FileExt"; 102 if (!$Options{overwrite}) { 103 if (-e $NewSDFile) { 104 warn "Warning: Ignoring file $SDFile: New SD file, $NewSDFile, already exists\n"; 105 next FILELIST; 106 } 107 if ($Options{keep}) { 108 if (-e $NewKeepSDFile) { 109 warn "Warning: Ignoring file $SDFile: New SD file, $NewKeepSDFile, already exists\n"; 110 next FILELIST; 111 } 112 } 113 } 114 if (lc($NewSDFile) eq lc($SDFile)) { 115 warn "Warning: Ignoring file $SDFile: Same output, $NewSDFile, and input file name\n"; 116 print "Specify a different name using \"-r --root\" option or use default name.\n"; 117 next FILELIST; 118 } 119 if (!open NEWSDFILE, ">$NewSDFile") { 120 warn "Warning: Ignoring file $SDFile: Couldn't open $NewSDFile: $! \n"; 121 next FILELIST; 122 } 123 if ($Options{keep}) { 124 if (!open NEWKEEPSDFILE, ">$NewKeepSDFile") { 125 warn "Warning: Ignoring file $SDFile: Couldn't open $NewKeepSDFile: $! \n"; 126 next FILELIST; 127 } 128 } 129 print "\nGenerating file $NewSDFile...\n"; 130 if ($Options{keep}) { 131 print "Generating file $NewKeepSDFile...\n"; 132 } 133 134 $CmpdCount = 0; $FilteredCmpdCount = 0; $KeepCmpdCount = 0; 135 $PrintCmpdCounterHeader = 1; 136 CMPDSTRING: while ($CmpdString = ReadCmpdString(\*SDFILE)) { 137 $CmpdCount++; 138 $FilterCmpd = 0; 139 if (($CmpdCount % 5000) == 0) { 140 if ($PrintCmpdCounterHeader) { 141 $PrintCmpdCounterHeader = 0; 142 print "\nProcessing compounds:"; 143 } 144 print "$CmpdCount..."; 145 } 146 @CmpdLines = split "\n", $CmpdString; 147 $CtabLinesCount = GetCtabLinesCount(\@CmpdLines); 148 if ($CtabLinesCount <= 0) { 149 $FilterCmpd = 1; 150 WriteOutCmpdString(); 151 next CMPDSTRING; 152 } 153 my ($AtomCount, $BondCount) = ParseCmpdCountsLine($CmpdLines[3]); 154 if ($Options{all} || $Options{mismatch}) { 155 if ($CtabLinesCount != ($AtomCount + $BondCount)) { 156 $FilterCmpd = 1; 157 WriteOutCmpdString(); 158 next CMPDSTRING; 159 } 160 } 161 if ($CtabLinesCount == ($AtomCount + $BondCount)) { 162 if ($Options{all} || $Options{unknownatoms}) { 163 my($UnknownAtomCount, $UnknownAtoms, $UnknownAtomLines) = GetUnknownAtoms(\@CmpdLines); 164 if ($UnknownAtomCount) { 165 $FilterCmpd = 1; 166 WriteOutCmpdString(); 167 next CMPDSTRING; 168 } 169 } 170 if ($Options{all} || $Options{cleansalts} || $Options{salts}) { 171 my ($FragmentsCount, $Fragments, $WashedCmpdString) = WashCmpd(\@CmpdLines); 172 if ($FragmentsCount > 1) { 173 if ($Options{all} || $Options{cleansalts}) { 174 $CmpdString = $WashedCmpdString; 175 } 176 else { 177 $FilterCmpd = 1; 178 } 179 WriteOutCmpdString(); 180 next CMPDSTRING; 181 } 182 } 183 } 184 WriteOutCmpdString(); 185 } 186 if (!$PrintCmpdCounterHeader) { 187 print "\n"; 188 } 189 close NEWSDFILE; 190 if ($Options{keep}) { 191 close NEWKEEPSDFILE; 192 } 193 close SDFILE; 194 print "\nTotal Number of compounds: $CmpdCount\n"; 195 print "Number of compounds left after filtering: $FilteredCmpdCount\n"; 196 print "Number of compounds ignored: $KeepCmpdCount\n"; 197 } 198 199 print "\n$ScriptName:Done...\n\n"; 200 201 $EndTime = new Benchmark; 202 $TotalTime = timediff ($EndTime, $StartTime); 203 print "Total time: ", timestr($TotalTime), "\n"; 204 205 ############################################################################### 206 207 # Write out the compound data... 208 sub WriteOutCmpdString { 209 if ($FilterCmpd) { 210 $KeepCmpdCount++; 211 if ($Options{keep}) { 212 print NEWKEEPSDFILE "$CmpdString\n"; 213 } 214 } 215 else { 216 $FilteredCmpdCount++; 217 print NEWSDFILE "$CmpdString\n"; 218 } 219 } 220 221 # Setup script usage and retrieve command line arguments specified using various options... 222 sub SetupScriptUsage { 223 224 # Retrieve all the options... 225 %Options = (); 226 if (!GetOptions(\%Options, "all|a", "cleansalts|c", "empty|e", "help|h", "keep|k", "mismatch|m", "overwrite|o", "root|r=s", "salts|s", "unknownatoms|u", "workingdir|w=s")) { 227 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"; 228 } 229 if ($Options{workingdir}) { 230 if (! -d $Options{workingdir}) { 231 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"; 232 } 233 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n"; 234 } 235 } 236
