1 #!/usr/bin/perl -w
   2 #
   3 # $RCSfile: AtomTypesFingerprints.pl,v $
   4 # $Date: 2012/03/17 21:35:08 $
   5 # $Revision: 1.18 $
   6 #
   7 # Author: Manish Sud <msud@san.rr.com>
   8 #
   9 # Copyright (C) 2004-2012 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 strict;
  30 use FindBin; use lib "$FindBin::Bin/../lib";
  31 use Getopt::Long;
  32 use File::Basename;
  33 use Text::ParseWords;
  34 use Benchmark;
  35 use FileUtil;
  36 use TextUtil;
  37 use SDFileUtil;
  38 use MoleculeFileIO;
  39 use FileIO::FingerprintsSDFileIO;
  40 use FileIO::FingerprintsTextFileIO;
  41 use FileIO::FingerprintsFPFileIO;
  42 use AtomTypes::AtomicInvariantsAtomTypes;
  43 use AtomTypes::FunctionalClassAtomTypes;
  44 use Fingerprints::AtomTypesFingerprints;
  45 
  46 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
  47 
  48 # Autoflush STDOUT
  49 $| = 1;
  50 
  51 # Starting message...
  52 $ScriptName = basename($0);
  53 print "\n$ScriptName: Starting...\n\n";
  54 $StartTime = new Benchmark;
  55 
  56 # Get the options and setup script...
  57 SetupScriptUsage();
  58 if ($Options{help} || @ARGV < 1) {
  59   die GetUsageFromPod("$FindBin::Bin/$ScriptName");
  60 }
  61 
  62 my(@SDFilesList);
  63 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd");
  64 
  65 # Process options...
  66 print "Processing options...\n";
  67 my(%OptionsInfo);
  68 ProcessOptions();
  69 
  70 # Setup information about input files...
  71 print "Checking input SD file(s)...\n";
  72 my(%SDFilesInfo);
  73 RetrieveSDFilesInfo();
  74 
  75 # Process input files..
  76 my($FileIndex);
  77 if (@SDFilesList > 1) {
  78   print "\nProcessing SD files...\n";
  79 }
  80 for $FileIndex (0 .. $#SDFilesList) {
  81   if ($SDFilesInfo{FileOkay}[$FileIndex]) {
  82     print "\nProcessing file $SDFilesList[$FileIndex]...\n";
  83     GenerateAtomTypesFingerprints($FileIndex);
  84   }
  85 }
  86 print "\n$ScriptName:Done...\n\n";
  87 
  88 $EndTime = new Benchmark;
  89 $TotalTime = timediff ($EndTime, $StartTime);
  90 print "Total time: ", timestr($TotalTime), "\n";
  91 
  92 ###############################################################################
  93 
  94 # Generate fingerprints for a SD file...
  95 #
  96 sub GenerateAtomTypesFingerprints {
  97   my($FileIndex) = @_;
  98   my($CmpdCount, $IgnoredCmpdCount, $SDFile, $MoleculeFileIO, $Molecule, $AtomTypesFingerprints, $NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO);
  99 
 100   $SDFile = $SDFilesList[$FileIndex];
 101 
 102   # Setup output files...
 103   #
 104   ($NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO) = SetupAndOpenOutputFiles($FileIndex);
 105 
 106   $MoleculeFileIO = new MoleculeFileIO('Name' => $SDFile);
 107   $MoleculeFileIO->Open();
 108 
 109   $CmpdCount = 0;
 110   $IgnoredCmpdCount = 0;
 111 
 112   COMPOUND: while ($Molecule = $MoleculeFileIO->ReadMolecule()) {
 113     $CmpdCount++;
 114 
 115     # Filter compound data before calculating fingerprints...
 116     if ($OptionsInfo{Filter}) {
 117       if (CheckAndFilterCompound($CmpdCount, $Molecule)) {
 118         $IgnoredCmpdCount++;
 119         next COMPOUND;
 120       }
 121     }
 122 
 123     $AtomTypesFingerprints = GenerateMoleculeFingerprints($Molecule);
 124     if (!$AtomTypesFingerprints) {
 125       $IgnoredCmpdCount++;
 126       ProcessIgnoredCompound('FingerprintsGenerationFailed', $CmpdCount, $Molecule);
 127       next COMPOUND;
 128     }
 129 
 130     WriteDataToOutputFiles($FileIndex, $CmpdCount, $Molecule, $AtomTypesFingerprints, $NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO);
 131   }
 132   $MoleculeFileIO->Close();
 133 
 134   if ($NewFPSDFileIO) {
 135     $NewFPSDFileIO->Close();
 136   }
 137   if ($NewFPTextFileIO) {
 138     $NewFPTextFileIO->Close();
 139   }
 140   if ($NewFPFileIO) {
 141     $NewFPFileIO->Close();
 142   }
 143 
 144   WriteFingerprintsGenerationSummaryStatistics($CmpdCount, $IgnoredCmpdCount);
 145 }
 146 
 147 # Process compound being ignored due to problems in fingerprints geneation...
 148 #
 149 sub ProcessIgnoredCompound {
 150   my($Mode, $CmpdCount, $Molecule) = @_;
 151   my($CmpdID, $DataFieldLabelAndValuesRef);
 152 
 153   $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues();
 154   $CmpdID = SetupCmpdIDForOutputFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
 155 
 156   MODE: {
 157     if ($Mode =~ /^ContainsNonElementalData$/i) {
 158       warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains atom data corresponding to non-elemental atom symbol(s)...\n\n";
 159       next MODE;
 160     }
 161 
 162     if ($Mode =~ /^ContainsNoElementalData$/i) {
 163       warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains no atom data...\n\n";
 164       next MODE;
 165     }
 166 
 167     if ($Mode =~ /^FingerprintsGenerationFailed$/i) {
 168       warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn't succeed...\n\n";
 169       next MODE;
 170     }
 171     warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Fingerprints generation didn't succeed...\n\n";
 172   }
 173 }
 174 
 175 # Check and filter compounds....
 176 #
 177 sub CheckAndFilterCompound {
 178   my($CmpdCount, $Molecule) = @_;
 179   my($ElementCount, $NonElementCount);
 180 
 181   ($ElementCount, $NonElementCount) = $Molecule->GetNumOfElementsAndNonElements();
 182 
 183   if ($NonElementCount) {
 184     ProcessIgnoredCompound('ContainsNonElementalData', $CmpdCount, $Molecule);
 185     return 1;
 186   }
 187 
 188   if (!$ElementCount) {
 189     ProcessIgnoredCompound('ContainsNoElementalData', $CmpdCount, $Molecule);
 190     return 1;
 191   }
 192 
 193   return 0;
 194 }
 195 
 196 # Write out compounds fingerprints generation summary statistics...
 197 #
 198 sub WriteFingerprintsGenerationSummaryStatistics {
 199   my($CmpdCount, $IgnoredCmpdCount) = @_;
 200   my($ProcessedCmpdCount);
 201 
 202   $ProcessedCmpdCount = $CmpdCount - $IgnoredCmpdCount;
 203 
 204   print "\nNumber of compounds: $CmpdCount\n";
 205   print "Number of compounds processed successfully during fingerprints generation: $ProcessedCmpdCount\n";
 206   print "Number of compounds ignored during fingerprints generation: $IgnoredCmpdCount\n";
 207 }
 208 
 209 # Open output files...
 210 #
 211 sub SetupAndOpenOutputFiles {
 212   my($FileIndex) = @_;
 213   my($NewFPSDFile, $NewFPFile, $NewFPTextFile, $NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO, %FingerprintsFileIOParams);
 214 
 215   ($NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO) = (undef) x 3;
 216 
 217   # Setup common parameters for fingerprints file IO objects...
 218   #
 219   %FingerprintsFileIOParams = ();
 220   if ($OptionsInfo{Mode} =~ /^AtomTypesBits$/i) {
 221     %FingerprintsFileIOParams = ('Mode' => 'Write', 'Overwrite' => $OptionsInfo{OverwriteFiles}, 'FingerprintsStringMode' => 'FingerprintsBitVectorString', 'BitStringFormat' => $OptionsInfo{BitStringFormat}, 'BitsOrder' => $OptionsInfo{BitsOrder});
 222   }
 223   elsif ($OptionsInfo{Mode} =~ /^AtomTypesCount$/i) {
 224     %FingerprintsFileIOParams = ('Mode' => 'Write', 'Overwrite' => $OptionsInfo{OverwriteFiles}, 'FingerprintsStringMode' => 'FingerprintsVectorString', 'VectorStringFormat' => $OptionsInfo{VectorStringFormat});
 225   }
 226 
 227   if ($OptionsInfo{SDOutput}) {
 228     $NewFPSDFile = $SDFilesInfo{SDOutFileNames}[$FileIndex];
 229     print "Generating SD file $NewFPSDFile...\n";
 230     $NewFPSDFileIO = new FingerprintsSDFileIO('Name' => $NewFPSDFile, %FingerprintsFileIOParams, 'FingerprintsFieldLabel' => $OptionsInfo{FingerprintsLabel});
 231     $NewFPSDFileIO->Open();
 232   }
 233 
 234   if ($OptionsInfo{FPOutput}) {
 235     $NewFPFile = $SDFilesInfo{FPOutFileNames}[$FileIndex];
 236     print "Generating FP file $NewFPFile...\n";
 237     $NewFPFileIO = new FingerprintsFPFileIO('Name' => $NewFPFile, %FingerprintsFileIOParams);
 238     $NewFPFileIO->Open();
 239   }
 240 
 241   if ($OptionsInfo{TextOutput}) {
 242     my($ColLabelsRef);
 243 
 244     $NewFPTextFile = $SDFilesInfo{TextOutFileNames}[$FileIndex];
 245     $ColLabelsRef = SetupFPTextFileCoulmnLabels($FileIndex);
 246 
 247     print "Generating text file $NewFPTextFile...\n";
 248     $NewFPTextFileIO = new FingerprintsTextFileIO('Name' => $NewFPTextFile, %FingerprintsFileIOParams, 'DataColLabels' => $ColLabelsRef, 'OutDelim' => $OptionsInfo{OutDelim}, 'OutQuote' => $OptionsInfo{OutQuote});
 249     $NewFPTextFileIO->Open();
 250   }
 251 
 252   return ($NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO);
 253 }
 254 
 255 # Write fingerpritns and other data to appropriate output files...
 256 #
 257 sub WriteDataToOutputFiles {
 258   my($FileIndex, $CmpdCount, $Molecule, $AtomTypesFingerprints, $NewFPSDFileIO, $NewFPTextFileIO, $NewFPFileIO) = @_;
 259   my($DataFieldLabelAndValuesRef);
 260 
 261   $DataFieldLabelAndValuesRef = undef;
 262   if ($NewFPTextFileIO || $NewFPFileIO) {
 263     $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues();
 264   }
 265 
 266   if ($NewFPSDFileIO) {
 267     my($CmpdString);
 268 
 269     $CmpdString = $Molecule->GetInputMoleculeString();
 270     $NewFPSDFileIO->WriteFingerprints($AtomTypesFingerprints, $CmpdString);
 271   }
 272 
 273   if ($NewFPTextFileIO) {
 274     my($ColValuesRef);
 275 
 276     $ColValuesRef = SetupFPTextFileCoulmnValues($FileIndex, $CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
 277     $NewFPTextFileIO->WriteFingerprints($AtomTypesFingerprints, $ColValuesRef);
 278   }
 279 
 280   if ($NewFPFileIO) {
 281     my($CompoundID);
 282 
 283     $CompoundID = SetupCmpdIDForOutputFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
 284     $NewFPFileIO->WriteFingerprints($AtomTypesFingerprints, $CompoundID);
 285   }
 286 }
 287 
 288 # Generate approriate column labels for FPText output file...
 289 #
 290 sub SetupFPTextFileCoulmnLabels {
 291   my($FileIndex) = @_;
 292   my($Line, @ColLabels);
 293 
 294   @ColLabels = ();
 295   if ($OptionsInfo{DataFieldsMode} =~ /^All$/i) {
 296     push @ColLabels, @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]};
 297   }
 298   elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) {
 299     push @ColLabels, @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]};
 300   }
 301   elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) {
 302     push @ColLabels, @{$OptionsInfo{SpecifiedDataFields}};
 303   }
 304   elsif ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) {
 305     push @ColLabels, $OptionsInfo{CompoundIDLabel};
 306   }
 307   # Add fingerprints label...
 308   push @ColLabels, $OptionsInfo{FingerprintsLabel};
 309 
 310   return \@ColLabels;
 311 }
 312 
 313 # Generate column values FPText output file..
 314 #
 315 sub SetupFPTextFileCoulmnValues {
 316   my($FileIndex, $CmpdCount, $Molecule, $DataFieldLabelAndValuesRef) = @_;
 317   my(@ColValues);
 318 
 319   @ColValues = ();
 320   if ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) {
 321     push @ColValues, SetupCmpdIDForOutputFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef);
 322   }
 323   elsif ($OptionsInfo{DataFieldsMode} =~ /^All$/i) {
 324     @ColValues = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]};
 325   }
 326   elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) {
 327     @ColValues = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]};
 328   }
 329   elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) {
 330     @ColValues = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$OptionsInfo{SpecifiedDataFields}};
 331   }
 332 
 333   return \@ColValues;
 334 }
 335 
 336 # Generate compound ID for FP and FPText output files..
 337 #
 338 sub SetupCmpdIDForOutputFiles {
 339   my($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef) = @_;
 340   my($CmpdID);
 341 
 342   $CmpdID = '';
 343   if ($OptionsInfo{CompoundIDMode} =~ /^MolNameOrLabelPrefix$/i) {
 344     my($MolName);
 345     $MolName = $Molecule->GetName();
 346     $CmpdID = $MolName ? $MolName : "$OptionsInfo{CompoundID}${CmpdCount}";
 347   }
 348   elsif ($OptionsInfo{CompoundIDMode} =~ /^LabelPrefix$/i) {
 349     $CmpdID = "$OptionsInfo{CompoundID}${CmpdCount}";
 350   }
 351   elsif ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i) {
 352     my($SpecifiedDataField);
 353     $SpecifiedDataField = $OptionsInfo{CompoundID};
 354     $CmpdID = exists $DataFieldLabelAndValuesRef->{$SpecifiedDataField} ? $DataFieldLabelAndValuesRef->{$SpecifiedDataField} : '';
 355   }
 356   elsif ($OptionsInfo{CompoundIDMode} =~ /^MolName$/i) {
 357     $CmpdID = $Molecule->GetName();
 358   }
 359   return $CmpdID;
 360 }
 361 
 362 # Generate fingerprints for molecule...
 363 #
 364 sub GenerateMoleculeFingerprints {
 365   my($Molecule) = @_;
 366   my($AtomTypesFingerprints);
 367 
 368   if ($OptionsInfo{KeepLargestComponent}) {
 369     $Molecule->KeepLargestComponent();
 370   }
 371   if (!$Molecule->DetectRings()) {
 372     return undef;
 373   }
 374   $Molecule->DetectAromaticity();
 375 
 376   $AtomTypesFingerprints = undef;
 377   if ($OptionsInfo{Mode} =~ /^AtomTypesCount$/i) {
 378     $AtomTypesFingerprints = new AtomTypesFingerprints('Molecule' => $Molecule, 'Type' => 'AtomTypesCount', 'AtomIdentifierType' => $OptionsInfo{AtomIdentifierType}, 'AtomTypesSetToUse' => $OptionsInfo{AtomTypesSetToUse}, 'IgnoreHydrogens' => $OptionsInfo{IgnoreHydrogens});
 379 
 380   }
 381   elsif ($OptionsInfo{Mode} =~ /^AtomTypesBits$/i) {
 382     $AtomTypesFingerprints = new AtomTypesFingerprints('Molecule' => $Molecule, 'Type' => 'AtomTypesBits', 'AtomIdentifierType' => $OptionsInfo{AtomIdentifierType}, 'AtomTypesSetToUse' => 'FixedSize', 'IgnoreHydrogens' => $OptionsInfo{IgnoreHydrogens});
 383   }
 384   else {
 385     die "Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: AtomTypesCount or AtomTypesBits\n";
 386   }
 387 
 388   SetAtomIdentifierTypeValuesToUse($AtomTypesFingerprints);
 389 
 390   # Generate atom types fingerprints...
 391   $AtomTypesFingerprints->GenerateFingerprints();
 392 
 393   # Make sure atom types fingerprints generation is successful...
 394   if (!$AtomTypesFingerprints->IsFingerprintsGenerationSuccessful()) {
 395     return undef;
 396   }
 397 
 398   return $AtomTypesFingerprints;
 399 }
 400 
 401 # Set atom identifier type to use for generating fingerprints...
 402 #
 403 sub SetAtomIdentifierTypeValuesToUse {
 404   my($AtomTypesFingerprints) = @_;
 405 
 406   if ($OptionsInfo{AtomIdentifierType} =~ /^AtomicInvariantsAtomTypes$/i) {
 407     $AtomTypesFingerprints->SetAtomicInvariantsToUse(\@{$OptionsInfo{AtomicInvariantsToUse}});
 408   }
 409   elsif ($OptionsInfo{AtomIdentifierType} =~ /^FunctionalClassAtomTypes$/i) {
 410     $AtomTypesFingerprints->SetFunctionalClassesToUse(\@{$OptionsInfo{FunctionalClassesToUse}});
 411   }
 412   elsif ($OptionsInfo{AtomIdentifierType} =~ /^(DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i) {
 413     # Nothing to do for now...
 414   }
 415   else {
 416     die "Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n";
 417   }
 418 }
 419 
 420 # Retrieve information about SD files...
 421 #
 422 sub RetrieveSDFilesInfo {
 423   my($SDFile, $Index, $FileDir, $FileExt, $FileName, $OutFileRoot, $TextOutFileExt, $SDOutFileExt, $FPOutFileExt, $NewSDFileName, $NewFPFileName, $NewTextFileName, $CheckDataField, $CollectDataFields, $AllDataFieldsRef, $CommonDataFieldsRef);
 424 
 425   %SDFilesInfo = ();
 426   @{$SDFilesInfo{FileOkay}} = ();
 427   @{$SDFilesInfo{OutFileRoot}} = ();
 428   @{$SDFilesInfo{SDOutFileNames}} = ();
 429   @{$SDFilesInfo{FPOutFileNames}} = ();
 430   @{$SDFilesInfo{TextOutFileNames}} = ();
 431   @{$SDFilesInfo{AllDataFieldsRef}} = ();
 432   @{$SDFilesInfo{CommonDataFieldsRef}} = ();
 433 
 434   $CheckDataField = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) && ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i)) ? 1 : 0;
 435   $CollectDataFields = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^(All|Common)$/i)) ? 1 : 0;
 436 
 437   FILELIST: for $Index (0 .. $#SDFilesList) {
 438     $SDFile = $SDFilesList[$Index];
 439 
 440     $SDFilesInfo{FileOkay}[$Index] = 0;
 441     $SDFilesInfo{OutFileRoot}[$Index] = '';
 442     $SDFilesInfo{SDOutFileNames}[$Index] = '';
 443     $SDFilesInfo{FPOutFileNames}[$Index] = '';
 444     $SDFilesInfo{TextOutFileNames}[$Index] = '';
 445 
 446     $SDFile = $SDFilesList[$Index];
 447     if (!(-e $SDFile)) {
 448       warn "Warning: Ignoring file $SDFile: It doesn't exist\n";
 449       next FILELIST;
 450     }
 451     if (!CheckFileType($SDFile, "sd sdf")) {
 452       warn "Warning: Ignoring file $SDFile: It's not a SD file\n";
 453       next FILELIST;
 454     }
 455 
 456     if ($CheckDataField) {
 457       # Make sure data field exists in SD file..
 458       my($CmpdString, $SpecifiedDataField, @CmpdLines, %DataFieldValues);
 459 
 460       @CmpdLines = ();
 461       open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n";
 462       $CmpdString = ReadCmpdString(\*SDFILE);
 463       close SDFILE;
 464       @CmpdLines = split "\n", $CmpdString;
 465       %DataFieldValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines);
 466       $SpecifiedDataField = $OptionsInfo{CompoundID};
 467       if (!exists $DataFieldValues{$SpecifiedDataField}) {
 468         warn "Warning: Ignoring file $SDFile: Data field value, $SpecifiedDataField, using  \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\" doesn't exist\n";
 469         next FILELIST;
 470       }
 471     }
 472 
 473     $AllDataFieldsRef = '';
 474     $CommonDataFieldsRef = '';
 475     if ($CollectDataFields) {
 476       my($CmpdCount);
 477       open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n";
 478       ($CmpdCount, $AllDataFieldsRef, $CommonDataFieldsRef) = GetAllAndCommonCmpdDataHeaderLabels(\*SDFILE);
 479       close SDFILE;
 480     }
 481 
 482     # Setup output file names...
 483     $FileDir = ""; $FileName = ""; $FileExt = "";
 484     ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile);
 485 
 486     $TextOutFileExt = "csv";
 487     if ($Options{outdelim} =~ /^tab$/i) {
 488       $TextOutFileExt = "tsv";
 489     }
 490     $SDOutFileExt = $FileExt;
 491     $FPOutFileExt = "fpf";
 492 
 493     if ($OptionsInfo{OutFileRoot} && (@SDFilesList == 1)) {
 494       my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($OptionsInfo{OutFileRoot});
 495       if ($RootFileName && $RootFileExt) {
 496         $FileName = $RootFileName;
 497       }
 498       else {
 499         $FileName = $OptionsInfo{OutFileRoot};
 500       }
 501       $OutFileRoot = $FileName;
 502     }
 503     else {
 504       $OutFileRoot = $FileName . 'AtomTypesFP';
 505     }
 506 
 507     $NewSDFileName = "${OutFileRoot}.${SDOutFileExt}";
 508     $NewFPFileName = "${OutFileRoot}.${FPOutFileExt}";
 509     $NewTextFileName = "${OutFileRoot}.${TextOutFileExt}";
 510 
 511     if ($OptionsInfo{SDOutput}) {
 512       if ($SDFile =~ /$NewSDFileName/i) {
 513         warn "Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n";
 514         print "Specify a different name using \"-r --root\" option or use default name.\n";
 515         next FILELIST;
 516       }
 517     }
 518 
 519     if (!$OptionsInfo{OverwriteFiles}) {
 520       # Check SD and text outout files...
 521       if ($OptionsInfo{SDOutput}) {
 522         if (-e $NewSDFileName) {
 523           warn "Warning: Ignoring file $SDFile: The file $NewSDFileName already exists\n";
 524           next FILELIST;
 525         }
 526       }
 527       if ($OptionsInfo{FPOutput}) {
 528         if (-e $NewFPFileName) {
 529           warn "Warning: Ignoring file $SDFile: The file $NewFPFileName already exists\n";
 530           next FILELIST;
 531         }
 532       }
 533       if ($OptionsInfo{TextOutput}) {
 534         if (-e $NewTextFileName) {
 535           warn "Warning: Ignoring file $SDFile: The file $NewTextFileName already exists\n";
 536           next FILELIST;
 537         }
 538       }
 539     }
 540 
 541     $SDFilesInfo{FileOkay}[$Index] = 1;
 542 
 543     $SDFilesInfo{OutFileRoot}[$Index] = $OutFileRoot;
 544     $SDFilesInfo{SDOutFileNames}[$Index] = $NewSDFileName;
 545     $SDFilesInfo{FPOutFileNames}[$Index] = $NewFPFileName;
 546     $SDFilesInfo{TextOutFileNames}[$Index] = $NewTextFileName;
 547 
 548     $SDFilesInfo{AllDataFieldsRef}[$Index] = $AllDataFieldsRef;
 549     $SDFilesInfo{CommonDataFieldsRef}[$Index] = $CommonDataFieldsRef;
 550   }
 551 }
 552 
 553 # Process option values...
 554 sub ProcessOptions {
 555   %OptionsInfo = ();
 556 
 557   $OptionsInfo{Mode} = $Options{mode};
 558 
 559   ProcessAtomIdentifierTypeOptions();
 560 
 561   my($AtomTypesSetToUse);
 562   $AtomTypesSetToUse = '';
 563   if ($Options{mode} =~ /^AtomTypesBits$/i) {
 564     if ($Options{atomtypessettouse} && $Options{atomtypessettouse} !~ /^FixedSize$/) {
 565       die "Error: The value specified, $Options{atomtypessettouse}, for option \"-e, --AtomTypesSetToUse\" is not valid. Allowed values for AtomTypesBits of \"-m, --mode\" option: FixedSize\n";
 566     }
 567     $AtomTypesSetToUse = 'FixedSize';
 568   }
 569   else {
 570     if ($Options{atomidentifiertype} =~ /^(AtomicInvariantsAtomTypes|FunctionalClassAtomTypes)$/i && $Options{atomtypessettouse} =~ /^FixedSize$/) {
 571       die "Error: The value specified, $Options{atomtypessettouse}, for option \"-e, --AtomTypesSetToUse\" is not valid during \"AtomicInvariantsAtomTypes or FunctionalClassAtomTypes\" value of \"-a, --AtomIdentifierType\". Allowed values: ArbitrarySize\n";
 572     }
 573     if ($Options{atomidentifiertype} =~ /^TPSAAtomTypes$/i && $Options{atomtypessettouse} =~ /^ArbitrarySize$/) {
 574       die "Error: The value specified, $Options{atomtypessettouse}, for option \"-e, --AtomTypesSetToUse\" is not valid during \"TPSAAtomTypes\" value of \"-a, --AtomIdentifierType\". Allowed values: FixedSize\n";
 575     }
 576     $AtomTypesSetToUse = $Options{atomtypessettouse} ? $Options{atomtypessettouse} : 'ArbitrarySize';
 577   }
 578   $OptionsInfo{AtomTypesSetToUse} = $AtomTypesSetToUse;
 579 
 580   $OptionsInfo{BitsOrder} = $Options{bitsorder};
 581   $OptionsInfo{BitStringFormat} = $Options{bitstringformat};
 582 
 583   $OptionsInfo{CompoundIDMode} = $Options{compoundidmode};
 584   $OptionsInfo{CompoundIDLabel} = $Options{compoundidlabel};
 585   $OptionsInfo{DataFieldsMode} = $Options{datafieldsmode};
 586 
 587   my(@SpecifiedDataFields);
 588   @SpecifiedDataFields = ();
 589 
 590   @{$OptionsInfo{SpecifiedDataFields}} = ();
 591   $OptionsInfo{CompoundID} = '';
 592 
 593   if ($Options{datafieldsmode} =~ /^CompoundID$/i) {
 594     if ($Options{compoundidmode} =~ /^DataField$/i) {
 595       if (!$Options{compoundid}) {
 596         die "Error: You must specify a value for \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\". \n";
 597       }
 598       $OptionsInfo{CompoundID} = $Options{compoundid};
 599     }
 600     elsif ($Options{compoundidmode} =~ /^(LabelPrefix|MolNameOrLabelPrefix)$/i) {
 601       $OptionsInfo{CompoundID} = $Options{compoundid} ? $Options{compoundid} : 'Cmpd';
 602     }
 603   }
 604   elsif ($Options{datafieldsmode} =~ /^Specify$/i) {
 605     if (!$Options{datafields}) {
 606       die "Error: You must specify a value for \"--DataFields\" option in \"Specify\" \"-d, --DataFieldsMode\". \n";
 607     }
 608     @SpecifiedDataFields = split /\,/, $Options{datafields};
 609     push @{$OptionsInfo{SpecifiedDataFields}}, @SpecifiedDataFields;
 610   }
 611 
 612   $OptionsInfo{IgnoreHydrogens} = ($Options{ignorehydrogens} =~ /^Yes$/i) ? 1 : 0;
 613 
 614   $OptionsInfo{FingerprintsLabel} = $Options{fingerprintslabel} ? $Options{fingerprintslabel} : 'AtomTypesFingerprints';
 615 
 616   $OptionsInfo{Filter} = ($Options{filter} =~ /^Yes$/i) ? 1 : 0;
 617 
 618   if ($Options{fingerprintslabelmode} =~ /^FingerprintsLabelWithIDs$/) {
 619     if ($Options{mode} =~ /^(AtomTypesCount)$/i && $Options{atomtypessettouse} =~ /^FixedSize$/i) {
 620       # Append atom types to the fingerprints label...
 621       my($FixedSizeAtomTypesSetRef);
 622       $FixedSizeAtomTypesSetRef = GetFixedSizeAtomTypesSet();
 623 
 624       $OptionsInfo{FingerprintsLabel} .= "; AtomTypes: " . TextUtil::JoinWords($FixedSizeAtomTypesSetRef, " ", 0);
 625     }
 626   }
 627   $OptionsInfo{FingerprintsLabelMode} = $Options{fingerprintslabelmode};
 628 
 629   $OptionsInfo{KeepLargestComponent} = ($Options{keeplargestcomponent} =~ /^Yes$/i) ? 1 : 0;
 630 
 631   $OptionsInfo{Output} = $Options{output};
 632   $OptionsInfo{SDOutput} = ($Options{output} =~ /^(SD|All)$/i) ? 1 : 0;
 633   $OptionsInfo{FPOutput} = ($Options{output} =~ /^(FP|All)$/i) ? 1 : 0;
 634   $OptionsInfo{TextOutput} = ($Options{output} =~ /^(Text|All)$/i) ? 1 : 0;
 635 
 636   $OptionsInfo{OutDelim} = $Options{outdelim};
 637   $OptionsInfo{OutQuote} = ($Options{quote} =~ /^Yes$/i) ? 1 : 0;
 638 
 639   $OptionsInfo{OverwriteFiles} = $Options{overwrite} ? 1 : 0;
 640   $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : 0;
 641 
 642   # Setup default vector string format...
 643   my($VectorStringFormat);
 644   $VectorStringFormat = '';
 645   if ($Options{vectorstringformat}) {
 646     $VectorStringFormat = $Options{vectorstringformat};
 647   }
 648   else {
 649     $VectorStringFormat = ($Options{atomtypessettouse} =~ /^FixedSize$/) ? "ValuesString" : "IDsAndValuesString";
 650   }
 651   $OptionsInfo{VectorStringFormat} = $VectorStringFormat;
 652 }
 653 
 654 # Process atom identifier type and related options...
 655 #
 656 sub ProcessAtomIdentifierTypeOptions {
 657 
 658   $OptionsInfo{AtomIdentifierType} = $Options{atomidentifiertype};
 659 
 660   if ($Options{atomidentifiertype} =~ /^AtomicInvariantsAtomTypes$/i) {
 661     ProcessAtomicInvariantsToUseOption();
 662   }
 663   elsif ($Options{atomidentifiertype} =~ /^FunctionalClassAtomTypes$/i) {
 664     ProcessFunctionalClassesToUse();
 665   }
 666   elsif ($OptionsInfo{AtomIdentifierType} =~ /^(DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i) {
 667     # Nothing to do for now...
 668   }
 669   else {
 670     die "Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n";
 671   }
 672 }
 673 
 674 # Process specified atomic invariants to use...
 675 #
 676 sub ProcessAtomicInvariantsToUseOption {
 677   my($AtomicInvariant, $AtomSymbolSpecified, @AtomicInvariantsWords);
 678 
 679   @{$OptionsInfo{AtomicInvariantsToUse}} = ();
 680   if (IsEmpty($Options{atomicinvariantstouse})) {
 681     die "Error: Atomic invariants value specified using \"--AtomicInvariantsToUse\" option is empty\n";
 682   }
 683   $AtomSymbolSpecified = 0;
 684   @AtomicInvariantsWords = split /\,/, $Options{atomicinvariantstouse};
 685   for $AtomicInvariant (@AtomicInvariantsWords) {
 686     if (!AtomicInvariantsAtomTypes::IsAtomicInvariantAvailable($AtomicInvariant)) {
 687       die "Error: Atomic invariant specified, $AtomicInvariant, using \"--AtomicInvariantsToUse\" option is not valid...\n ";
 688     }
 689     if ($AtomicInvariant =~ /^(AS|AtomSymbol)$/i) {
 690       $AtomSymbolSpecified = 1;
 691     }
 692     push @{$OptionsInfo{AtomicInvariantsToUse}}, $AtomicInvariant;
 693   }
 694   if (!$AtomSymbolSpecified) {
 695     die "Error: Atomic invariant, AS or AtomSymbol, must be specified as using \"--AtomicInvariantsToUse\" option...\n ";
 696   }
 697 }
 698 
 699 # Process specified functional classes invariants to use...
 700 #
 701 sub ProcessFunctionalClassesToUse {
 702   my($FunctionalClass, @FunctionalClassesToUseWords);
 703 
 704   @{$OptionsInfo{FunctionalClassesToUse}} = ();
 705   if (IsEmpty($Options{functionalclassestouse})) {
 706     die "Error: Functional classes value specified using \"--FunctionalClassesToUse\" option is empty\n";
 707   }
 708   @FunctionalClassesToUseWords = split /\,/, $Options{functionalclassestouse};
 709   for $FunctionalClass (@FunctionalClassesToUseWords) {
 710     if (!FunctionalClassAtomTypes::IsFunctionalClassAvailable($FunctionalClass)) {
 711       die "Error: Functional class specified, $FunctionalClass, using \"--FunctionalClassesToUse\" option is not valid...\n ";
 712     }
 713     push @{$OptionsInfo{FunctionalClassesToUse}}, $FunctionalClass;
 714   }
 715 }
 716 
 717 # Get fixed size atom types set...
 718 #
 719 sub GetFixedSizeAtomTypesSet {
 720   my($AtomTypesRef);
 721 
 722   $AtomTypesRef = undef;
 723 
 724   IDENTIFIERTYPE: {
 725     if ($OptionsInfo{AtomIdentifierType} =~ /^DREIDINGAtomTypes$/i) {
 726       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? DREIDINGAtomTypes::GetAllPossibleDREIDINGNonHydrogenAtomTypes() : DREIDINGAtomTypes::GetAllPossibleDREIDINGAtomTypes();
 727       last IDENTIFIERTYPE;
 728     }
 729 
 730     if ($OptionsInfo{AtomIdentifierType} =~ /^EStateAtomTypes$/i) {
 731       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? EStateAtomTypes::GetAllPossibleEStateNonHydrogenAtomTypes() : EStateAtomTypes::GetAllPossibleEStateAtomTypes();
 732       last IDENTIFIERTYPE;
 733     }
 734 
 735     if ($OptionsInfo{AtomIdentifierType} =~ /^MMFF94AtomTypes$/i) {
 736       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? MMFF94AtomTypes::GetAllPossibleMMFF94NonHydrogenAtomTypes() : MMFF94AtomTypes::GetAllPossibleMMFF94AtomTypes();
 737       last IDENTIFIERTYPE;
 738     }
 739 
 740     if ($OptionsInfo{AtomIdentifierType} =~ /^SLogPAtomTypes$/i) {
 741       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? SLogPAtomTypes::GetAllPossibleSLogPNonHydrogenAtomTypes() : SLogPAtomTypes::GetAllPossibleSLogPAtomTypes();
 742       last IDENTIFIERTYPE;
 743     }
 744 
 745     if ($OptionsInfo{AtomIdentifierType} =~ /^SYBYLAtomTypes$/i) {
 746       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? SYBYLAtomTypes::GetAllPossibleSYBYLNonHydrogenAtomTypes() : SYBYLAtomTypes::GetAllPossibleSYBYLAtomTypes();
 747       last IDENTIFIERTYPE;
 748     }
 749 
 750     if ($OptionsInfo{AtomIdentifierType} =~ /^TPSAAtomTypes$/i) {
 751       $AtomTypesRef = TPSAAtomTypes::GetAllPossibleTPSAAtomTypes();
 752       last IDENTIFIERTYPE;
 753     }
 754 
 755     if ($OptionsInfo{AtomIdentifierType} =~ /^UFFAtomTypes$/i) {
 756       $AtomTypesRef = $OptionsInfo{IgnoreHydrogens} ? UFFAtomTypes::GetAllPossibleUFFNonHydrogenAtomTypes() : UFFAtomTypes::GetAllPossibleUFFAtomTypes();
 757       last IDENTIFIERTYPE;
 758     }
 759     die "Error: GetFixedSizeAtomTypesSet: Atom types set for atom indentifier type, $OptionsInfo{AtomIdentifierType}, is not available...";
 760   }
 761 
 762   return $AtomTypesRef;
 763 }
 764 
 765 # Setup script usage  and retrieve command line arguments specified using various options...
 766 sub SetupScriptUsage {
 767 
 768   # Retrieve all the options...
 769   %Options = ();
 770 
 771   $Options{atomidentifiertype} = 'AtomicInvariantsAtomTypes';
 772   $Options{atomicinvariantstouse} = 'AS,X,BO,H,FC';
 773   $Options{functionalclassestouse} = 'HBD,HBA,PI,NI,Ar,Hal';
 774 
 775   $Options{atomtypessettouse} = 'ArbitrarySize';
 776 
 777   $Options{bitsorder} = 'Ascending';
 778   $Options{bitstringformat} = 'BinaryString';
 779 
 780   $Options{compoundidmode} = 'LabelPrefix';
 781   $Options{compoundidlabel} = 'CompoundID';
 782   $Options{datafieldsmode} = 'CompoundID';
 783 
 784   $Options{filter} = 'Yes';
 785 
 786   $Options{fingerprintslabelmode} = 'FingerprintsLabelOnly';
 787   $Options{keeplargestcomponent} = 'Yes';
 788 
 789   $Options{mode} = 'AtomTypesCount';
 790 
 791   $Options{ignorehydrogens} = 'Yes';
 792 
 793   $Options{quote} = 'yes';
 794 
 795   $Options{output} = 'text';
 796   $Options{outdelim} = 'comma';
 797   $Options{quote} = 'yes';
 798 
 799   $Options{vectorstringformat} = '';
 800 
 801   if (!GetOptions(\%Options, "atomidentifiertype|a=s", "atomicinvariantstouse=s", "functionalclassestouse=s", "atomtypessettouse|e=s", "bitsorder=s", "bitstringformat|b=s", "compoundid=s", "compoundidlabel=s", "compoundidmode=s", "datafields=s", "datafieldsmode|d=s", "filter|f=s", "fingerprintslabelmode=s", "fingerprintslabel=s",  "help|h", "ignorehydrogens|i=s", "keeplargestcomponent|k=s", "mode|m=s", "outdelim=s", "output=s", "overwrite|o", "quote|q=s", "root|r=s", "vectorstringformat|v=s", "workingdir|w=s")) {
 802     die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
 803   }
 804   if ($Options{workingdir}) {
 805     if (! -d $Options{workingdir}) {
 806       die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
 807     }
 808     chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
 809   }
 810   if ($Options{atomidentifiertype} !~ /^(AtomicInvariantsAtomTypes|DREIDINGAtomTypes|EStateAtomTypes|FunctionalClassAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i) {
 811     die "Error: The value specified, $Options{atomidentifiertype}, for option \"-a, --AtomIdentifierType\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes\n";
 812   }
 813   if ($Options{atomtypessettouse} && $Options{atomtypessettouse} !~ /^(ArbitrarySize|FixedSize)$/) {
 814     die "Error: The value specified, $Options{atomtypessettouse}, for option \"--AtomTypesSetToUse\" is not valid. Allowed values: ArbitrarySize or FixedSize\n";
 815   }
 816   if ($Options{bitsorder} !~ /^(Ascending|Descending)$/i) {
 817     die "Error: The value specified, $Options{bitsorder}, for option \"--BitsOrder\" is not valid. Allowed values: Ascending or Descending\n";
 818   }
 819   if ($Options{bitstringformat} !~ /^(BinaryString|HexadecimalString)$/i) {
 820     die "Error: The value specified, $Options{bitstringformat}, for option \"-b, --bitstringformat\" is not valid. Allowed values: BinaryString or HexadecimalString\n";
 821   }
 822   if ($Options{compoundidmode} !~ /^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i) {
 823     die "Error: The value specified, $Options{compoundidmode}, for option \"--CompoundIDMode\" is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n";
 824   }
 825   if ($Options{datafieldsmode} !~ /^(All|Common|Specify|CompoundID)$/i) {
 826     die "Error: The value specified, $Options{datafieldsmode}, for option \"-d, --DataFieldsMode\" is not valid. Allowed values: All, Common, Specify or CompoundID\n";
 827   }
 828   if ($Options{filter} !~ /^(Yes|No)$/i) {
 829     die "Error: The value specified, $Options{filter}, for option \"-f, --Filter\" is not valid. Allowed values: Yes or No\n";
 830   }
 831   if ($Options{fingerprintslabelmode} !~ /^(FingerprintsLabelOnly|FingerprintsLabelWithIDs)$/i) {
 832     die "Error: The value specified, $Options{fingerprintslabelmode}, for option \"--FingerprintsLabelMode\" is not valid. Allowed values: FingerprintsLabelOnly or FingerprintsLabelWithIDs\n";
 833   }
 834   if ($Options{ignorehydrogens} !~ /^(Yes|No)$/i) {
 835     die "Error: The value specified, $Options{ignorehydrogens}, for option \"-i, --IgnoreHydrogens\" is not valid. Allowed values: Yes or No\n";
 836   }
 837   if ($Options{keeplargestcomponent} !~ /^(Yes|No)$/i) {
 838     die "Error: The value specified, $Options{keeplargestcomponent}, for option \"-k, --KeepLargestComponent\" is not valid. Allowed values: Yes or No\n";
 839   }
 840   if ($Options{mode} !~ /^(AtomTypesCount|AtomTypesBits)$/i) {
 841     die "Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: AtomTypesCount, or AtomTypesBits\n";
 842   }
 843   if ($Options{output} !~ /^(SD|FP|text|all)$/i) {
 844     die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: SD, FP, text, or all\n";
 845   }
 846   if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
 847     die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
 848   }
 849   if ($Options{quote} !~ /^(Yes|No)$/i) {
 850     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: Yes or No\n";
 851   }
 852   if ($Options{outdelim} =~ /semicolon/i && $Options{quote} =~ /^No$/i) {
 853     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not allowed with, semicolon value of \"--outdelim\" option: Fingerprints string use semicolon as delimiter for various data fields and must be quoted.\n";
 854   }
 855   if ($Options{vectorstringformat} && $Options{vectorstringformat} !~ /^(ValuesString|IDsAndValuesString|IDsAndValuesPairsString|ValuesAndIDsString|ValuesAndIDsPairsString)$/i) {
 856     die "Error: The value specified, $Options{vectorstringformat}, for option \"-v, --VectorStringFormat\" is not valid. Allowed values: ValuesString, IDsAndValuesString, IDsAndValuesPairsString, ValuesAndIDsString or ValuesAndIDsPairsString\n";
 857   }
 858 }
 859