Mercurial > repos > deepakjadmin > mayatool3_test2
comparison bin/CalculatePhysicochemicalProperties.pl @ 0:4816e4a8ae95 draft default tip
Uploaded
| author | deepakjadmin |
|---|---|
| date | Wed, 20 Jan 2016 09:23:18 -0500 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:4816e4a8ae95 |
|---|---|
| 1 #!/usr/bin/perl -w | |
| 2 # | |
| 3 # $RCSfile: CalculatePhysicochemicalProperties.pl,v $ | |
| 4 # $Date: 2015/02/28 20:46:19 $ | |
| 5 # $Revision: 1.20 $ | |
| 6 # | |
| 7 # Author: Manish Sud <msud@san.rr.com> | |
| 8 # | |
| 9 # Copyright (C) 2015 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 "$FindBin::Bin/../lib"; | |
| 31 use Getopt::Long; | |
| 32 use File::Basename; | |
| 33 use Text::ParseWords; | |
| 34 use Benchmark; | |
| 35 use FileUtil; | |
| 36 use TextUtil; | |
| 37 use SDFileUtil; | |
| 38 use MoleculeFileIO; | |
| 39 use Molecule; | |
| 40 use AtomTypes::AtomicInvariantsAtomTypes; | |
| 41 use AtomTypes::FunctionalClassAtomTypes; | |
| 42 use MolecularDescriptors::MolecularDescriptorsGenerator; | |
| 43 | |
| 44 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime); | |
| 45 | |
| 46 # Autoflush STDOUT | |
| 47 $| = 1; | |
| 48 | |
| 49 # Starting message... | |
| 50 $ScriptName = basename($0); | |
| 51 print "\n$ScriptName: Starting...\n\n"; | |
| 52 $StartTime = new Benchmark; | |
| 53 | |
| 54 # Get the options and setup script... | |
| 55 SetupScriptUsage(); | |
| 56 if ($Options{help} || @ARGV < 1) { | |
| 57 die GetUsageFromPod("$FindBin::Bin/$ScriptName"); | |
| 58 } | |
| 59 | |
| 60 my(@SDFilesList); | |
| 61 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd"); | |
| 62 | |
| 63 # Process options... | |
| 64 print "Processing options...\n"; | |
| 65 my(%OptionsInfo); | |
| 66 ProcessOptions(); | |
| 67 | |
| 68 # Setup information about input files... | |
| 69 print "Checking input SD file(s)...\n"; | |
| 70 my(%SDFilesInfo); | |
| 71 RetrieveSDFilesInfo(); | |
| 72 | |
| 73 # Process input files.. | |
| 74 my($FileIndex); | |
| 75 if (@SDFilesList > 1) { | |
| 76 print "\nProcessing SD files...\n"; | |
| 77 } | |
| 78 for $FileIndex (0 .. $#SDFilesList) { | |
| 79 if ($SDFilesInfo{FileOkay}[$FileIndex]) { | |
| 80 print "\nProcessing file $SDFilesList[$FileIndex]...\n"; | |
| 81 CalculatePhysicochemicalProperties($FileIndex); | |
| 82 } | |
| 83 } | |
| 84 print "\n$ScriptName:Done...\n\n"; | |
| 85 | |
| 86 $EndTime = new Benchmark; | |
| 87 $TotalTime = timediff ($EndTime, $StartTime); | |
| 88 print "Total time: ", timestr($TotalTime), "\n"; | |
| 89 | |
| 90 ############################################################################### | |
| 91 | |
| 92 # Calculate physicochemical properties for a SD file... | |
| 93 # | |
| 94 sub CalculatePhysicochemicalProperties { | |
| 95 my($FileIndex) = @_; | |
| 96 my($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount, $SDFile, $MoleculeFileIO, $Molecule, $MolecularDescriptorsGenerator, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef); | |
| 97 | |
| 98 $SDFile = $SDFilesList[$FileIndex]; | |
| 99 | |
| 100 # Setup output files... | |
| 101 $NewSDFileRef = ''; $NewTextFileRef = ''; | |
| 102 ($NewSDFileRef, $NewTextFileRef) = SetupAndOpenOutputFiles($FileIndex); | |
| 103 | |
| 104 # Setup molecular descriptor generator to calculate property values for specifed | |
| 105 # property names... | |
| 106 $MolecularDescriptorsGenerator = SetupMolecularDescriptorsGenerator(); | |
| 107 | |
| 108 ($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount) = ('0') x 4; | |
| 109 | |
| 110 $MoleculeFileIO = new MoleculeFileIO('Name' => $SDFile); | |
| 111 $MoleculeFileIO->Open(); | |
| 112 | |
| 113 COMPOUND: while ($Molecule = $MoleculeFileIO->ReadMolecule()) { | |
| 114 $CmpdCount++; | |
| 115 | |
| 116 # Filter compound data before calculating physiochemical properties... | |
| 117 if ($OptionsInfo{Filter}) { | |
| 118 if (CheckAndFilterCompound($CmpdCount, $Molecule)) { | |
| 119 $IgnoredCmpdCount++; | |
| 120 next COMPOUND; | |
| 121 } | |
| 122 } | |
| 123 | |
| 124 # Calculate properties... | |
| 125 $PhysicochemicalPropertiesDataRef = CalculateMoleculeProperties($MolecularDescriptorsGenerator, $Molecule); | |
| 126 | |
| 127 if (!defined($PhysicochemicalPropertiesDataRef)) { | |
| 128 $IgnoredCmpdCount++; | |
| 129 ProcessIgnoredCompound('PropertiesCalculationFailed', $CmpdCount, $Molecule); | |
| 130 next COMPOUND; | |
| 131 } | |
| 132 | |
| 133 # Calculate any rule violations... | |
| 134 if ($OptionsInfo{RuleOf5Violations} && $PhysicochemicalPropertiesDataRef->{RuleOf5Violations}) { | |
| 135 $RuleOf5ViolationsCount++; | |
| 136 } | |
| 137 | |
| 138 if ($OptionsInfo{RuleOf3Violations} && $PhysicochemicalPropertiesDataRef->{RuleOf3Violations}) { | |
| 139 $RuleOf3ViolationsCount++; | |
| 140 } | |
| 141 | |
| 142 # Write out calculate properties... | |
| 143 WriteDataToOutputFiles($FileIndex, $CmpdCount, $Molecule, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef); | |
| 144 } | |
| 145 $MoleculeFileIO->Close(); | |
| 146 | |
| 147 if ($OptionsInfo{SDOutput} && $NewSDFileRef) { | |
| 148 close $NewSDFileRef; | |
| 149 } | |
| 150 if ($OptionsInfo{TextOutput} && $NewTextFileRef) { | |
| 151 close $NewTextFileRef; | |
| 152 } | |
| 153 | |
| 154 WriteCalculationSummaryStatistics($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount); | |
| 155 } | |
| 156 | |
| 157 # Process compound being ignored due to problems in physicochemical properties calculation... | |
| 158 # | |
| 159 sub ProcessIgnoredCompound { | |
| 160 my($Mode, $CmpdCount, $Molecule) = @_; | |
| 161 my($CmpdID, $DataFieldLabelAndValuesRef); | |
| 162 | |
| 163 $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues(); | |
| 164 $CmpdID = SetupCmpdIDForTextFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef); | |
| 165 | |
| 166 MODE: { | |
| 167 if ($Mode =~ /^ContainsNonElementalData$/i) { | |
| 168 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains atom data corresponding to non-elemental atom symbol(s)...\n\n"; | |
| 169 next MODE; | |
| 170 } | |
| 171 | |
| 172 if ($Mode =~ /^ContainsNoElementalData$/i) { | |
| 173 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Compound contains no atom data...\n\n"; | |
| 174 next MODE; | |
| 175 } | |
| 176 | |
| 177 if ($Mode =~ /^PropertiesCalculationFailed$/i) { | |
| 178 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Physicochemical properties calculation didn't succeed...\n\n"; | |
| 179 next MODE; | |
| 180 } | |
| 181 warn "\nWarning: Ignoring compound record number $CmpdCount with ID $CmpdID: Physicochemical properties calculation didn't succeed...\n\n"; | |
| 182 } | |
| 183 } | |
| 184 | |
| 185 # Check and filter compounds.... | |
| 186 # | |
| 187 sub CheckAndFilterCompound { | |
| 188 my($CmpdCount, $Molecule) = @_; | |
| 189 my($ElementCount, $NonElementCount); | |
| 190 | |
| 191 ($ElementCount, $NonElementCount) = $Molecule->GetNumOfElementsAndNonElements(); | |
| 192 | |
| 193 if ($NonElementCount) { | |
| 194 ProcessIgnoredCompound('ContainsNonElementalData', $CmpdCount, $Molecule); | |
| 195 return 1; | |
| 196 } | |
| 197 | |
| 198 if (!$ElementCount) { | |
| 199 ProcessIgnoredCompound('ContainsNoElementalData', $CmpdCount, $Molecule); | |
| 200 return 1; | |
| 201 } | |
| 202 | |
| 203 return 0; | |
| 204 } | |
| 205 | |
| 206 # Write out compounds physicochemical properties calculation summary statistics... | |
| 207 # | |
| 208 sub WriteCalculationSummaryStatistics { | |
| 209 my($CmpdCount, $IgnoredCmpdCount, $RuleOf5ViolationsCount, $RuleOf3ViolationsCount) = @_; | |
| 210 my($ProcessedCmpdCount); | |
| 211 | |
| 212 $ProcessedCmpdCount = $CmpdCount - $IgnoredCmpdCount; | |
| 213 | |
| 214 print "\nNumber of compounds: $CmpdCount\n"; | |
| 215 print "Number of compounds processed successfully during physicochemical properties calculation: $ProcessedCmpdCount\n"; | |
| 216 print "Number of compounds ignored during physicochemical properties calculation: $IgnoredCmpdCount\n"; | |
| 217 | |
| 218 if ($OptionsInfo{RuleOf5Violations}) { | |
| 219 print "Number of compounds with one or more RuleOf5 violations: $RuleOf5ViolationsCount\n"; | |
| 220 } | |
| 221 | |
| 222 if ($OptionsInfo{RuleOf3Violations}) { | |
| 223 print "Number of compounds with one or more RuleOf3 violations: $RuleOf3ViolationsCount\n"; | |
| 224 } | |
| 225 | |
| 226 } | |
| 227 | |
| 228 # Open output files... | |
| 229 # | |
| 230 sub SetupAndOpenOutputFiles { | |
| 231 my($FileIndex) = @_; | |
| 232 my($NewSDFile, $NewTextFile, $NewSDFileRef, $NewTextFileRef); | |
| 233 | |
| 234 $NewSDFileRef = ''; | |
| 235 $NewTextFileRef = ''; | |
| 236 | |
| 237 if ($OptionsInfo{SDOutput}) { | |
| 238 $NewSDFile = $SDFilesInfo{SDOutFileNames}[$FileIndex]; | |
| 239 print "Generating SD file $NewSDFile...\n"; | |
| 240 open NEWSDFILE, ">$NewSDFile" or die "Error: Couldn't open $NewSDFile: $! \n"; | |
| 241 $NewSDFileRef = \*NEWSDFILE; | |
| 242 } | |
| 243 if ($OptionsInfo{TextOutput}) { | |
| 244 $NewTextFile = $SDFilesInfo{TextOutFileNames}[$FileIndex]; | |
| 245 print "Generating text file $NewTextFile...\n"; | |
| 246 open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n"; | |
| 247 WriteTextFileCoulmnLabels($FileIndex, \*NEWTEXTFILE); | |
| 248 $NewTextFileRef = \*NEWTEXTFILE; | |
| 249 } | |
| 250 return ($NewSDFileRef, $NewTextFileRef); | |
| 251 } | |
| 252 | |
| 253 # Write calculated physicochemical properties and other data to appropriate output files... | |
| 254 # | |
| 255 sub WriteDataToOutputFiles { | |
| 256 my($FileIndex, $CmpdCount, $Molecule, $PhysicochemicalPropertiesDataRef, $NewSDFileRef, $NewTextFileRef) = @_; | |
| 257 my($PropertyName, $PropertyValue); | |
| 258 | |
| 259 if ($OptionsInfo{SDOutput}) { | |
| 260 # Retrieve input compound string used to create molecule and write it out | |
| 261 # without last line containing a delimiter... | |
| 262 my($CmpdString); | |
| 263 $CmpdString = $Molecule->GetInputMoleculeString(); | |
| 264 $CmpdString =~ s/\$\$\$\$$//; | |
| 265 print $NewSDFileRef "$CmpdString"; | |
| 266 | |
| 267 # Write out calculated physicochemical properties data... | |
| 268 for $PropertyName (@{$OptionsInfo{SpecifiedPropertyNames}}) { | |
| 269 $PropertyValue = $PhysicochemicalPropertiesDataRef->{$PropertyName}; | |
| 270 print $NewSDFileRef "> <$PropertyName>\n$PropertyValue\n\n"; | |
| 271 } | |
| 272 | |
| 273 # Write out RuleOf5 violations for molecule.... | |
| 274 if ($OptionsInfo{RuleOf5Violations}) { | |
| 275 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf5Violations}; | |
| 276 print $NewSDFileRef "> <RuleOf5Violations>\n$PropertyValue\n\n"; | |
| 277 } | |
| 278 | |
| 279 # Write out RuleOf3 violations for molecule.... | |
| 280 if ($OptionsInfo{RuleOf3Violations}) { | |
| 281 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf3Violations}; | |
| 282 print $NewSDFileRef "> <RuleOf3Violations>\n$PropertyValue\n\n"; | |
| 283 } | |
| 284 | |
| 285 # Write out delimiter... | |
| 286 print $NewSDFileRef "\$\$\$\$\n"; | |
| 287 } | |
| 288 | |
| 289 if ($OptionsInfo{TextOutput}) { | |
| 290 my($Line, $DataFieldLabelAndValuesRef, $DataFieldLabel, $DataFieldValue, @LineWords,); | |
| 291 | |
| 292 $DataFieldLabelAndValuesRef = $Molecule->GetDataFieldLabelAndValues(); | |
| 293 @LineWords = (); | |
| 294 if ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) { | |
| 295 push @LineWords, SetupCmpdIDForTextFiles($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef); | |
| 296 } | |
| 297 elsif ($OptionsInfo{DataFieldsMode} =~ /^All$/i) { | |
| 298 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]}; | |
| 299 } | |
| 300 elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) { | |
| 301 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]}; | |
| 302 } | |
| 303 elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) { | |
| 304 @LineWords = map { exists $DataFieldLabelAndValuesRef->{$_} ? $DataFieldLabelAndValuesRef->{$_} : ''} @{$OptionsInfo{SpecifiedDataFields}}; | |
| 305 } | |
| 306 | |
| 307 # Append calculated physicochemical properties data... | |
| 308 for $PropertyName (@{$OptionsInfo{SpecifiedPropertyNames}}) { | |
| 309 $PropertyValue = $PhysicochemicalPropertiesDataRef->{$PropertyName}; | |
| 310 push @LineWords, $PropertyValue; | |
| 311 } | |
| 312 | |
| 313 # Write out RuleOf5 violations for molecule.... | |
| 314 if ($OptionsInfo{RuleOf5Violations}) { | |
| 315 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf5Violations}; | |
| 316 push @LineWords, $PropertyValue; | |
| 317 } | |
| 318 | |
| 319 # Write out RuleOf3 violations for molecule.... | |
| 320 if ($OptionsInfo{RuleOf3Violations}) { | |
| 321 $PropertyValue = $PhysicochemicalPropertiesDataRef->{RuleOf3Violations}; | |
| 322 push @LineWords, $PropertyValue; | |
| 323 } | |
| 324 | |
| 325 $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote}); | |
| 326 print $NewTextFileRef "$Line\n"; | |
| 327 } | |
| 328 } | |
| 329 | |
| 330 # Write out approriate column labels to text file... | |
| 331 sub WriteTextFileCoulmnLabels { | |
| 332 my($FileIndex, $NewTextFileRef) = @_; | |
| 333 my($Line, @LineWords); | |
| 334 | |
| 335 @LineWords = (); | |
| 336 if ($OptionsInfo{DataFieldsMode} =~ /^All$/i) { | |
| 337 push @LineWords, @{$SDFilesInfo{AllDataFieldsRef}[$FileIndex]}; | |
| 338 } | |
| 339 elsif ($OptionsInfo{DataFieldsMode} =~ /^Common$/i) { | |
| 340 push @LineWords, @{$SDFilesInfo{CommonDataFieldsRef}[$FileIndex]}; | |
| 341 } | |
| 342 elsif ($OptionsInfo{DataFieldsMode} =~ /^Specify$/i) { | |
| 343 push @LineWords, @{$OptionsInfo{SpecifiedDataFields}}; | |
| 344 } | |
| 345 elsif ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) { | |
| 346 push @LineWords, $OptionsInfo{CompoundIDLabel}; | |
| 347 } | |
| 348 my($SpecifiedPropertyName); | |
| 349 | |
| 350 # Append physicochemical properties column labels... | |
| 351 push @LineWords, @{$OptionsInfo{SpecifiedPropertyNames}}; | |
| 352 | |
| 353 # Write out RuleOf5 violations label... | |
| 354 if ($OptionsInfo{RuleOf5Violations}) { | |
| 355 push @LineWords, 'RuleOf5Violations'; | |
| 356 } | |
| 357 | |
| 358 # Write out RuleOf3 violations label... | |
| 359 if ($OptionsInfo{RuleOf3Violations}) { | |
| 360 push @LineWords, 'RuleOf3Violations'; | |
| 361 } | |
| 362 | |
| 363 $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote}); | |
| 364 print $NewTextFileRef "$Line\n"; | |
| 365 } | |
| 366 | |
| 367 # Generate compound ID for text files.. | |
| 368 # | |
| 369 sub SetupCmpdIDForTextFiles { | |
| 370 my($CmpdCount, $Molecule, $DataFieldLabelAndValuesRef) = @_; | |
| 371 my($CmpdID); | |
| 372 | |
| 373 $CmpdID = ''; | |
| 374 if ($OptionsInfo{CompoundIDMode} =~ /^MolNameOrLabelPrefix$/i) { | |
| 375 my($MolName); | |
| 376 $MolName = $Molecule->GetName(); | |
| 377 $CmpdID = $MolName ? $MolName : "$OptionsInfo{CompoundID}${CmpdCount}"; | |
| 378 } | |
| 379 elsif ($OptionsInfo{CompoundIDMode} =~ /^LabelPrefix$/i) { | |
| 380 $CmpdID = "$OptionsInfo{CompoundID}${CmpdCount}"; | |
| 381 } | |
| 382 elsif ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i) { | |
| 383 my($SpecifiedDataField); | |
| 384 $SpecifiedDataField = $OptionsInfo{CompoundID}; | |
| 385 $CmpdID = exists $DataFieldLabelAndValuesRef->{$SpecifiedDataField} ? $DataFieldLabelAndValuesRef->{$SpecifiedDataField} : ''; | |
| 386 } | |
| 387 elsif ($OptionsInfo{CompoundIDMode} =~ /^MolName$/i) { | |
| 388 $CmpdID = $Molecule->GetName(); | |
| 389 } | |
| 390 return $CmpdID; | |
| 391 } | |
| 392 | |
| 393 # Calculate physicochemical properties for molecule... | |
| 394 # | |
| 395 sub CalculateMoleculeProperties { | |
| 396 my($MolecularDescriptorsGenerator, $Molecule) = @_; | |
| 397 my($PropertyName, $PropertyValue, $MolecularDescriptorsObject, %CalculatedPhysicochemicalProperties); | |
| 398 | |
| 399 %CalculatedPhysicochemicalProperties = (); | |
| 400 | |
| 401 if ($OptionsInfo{KeepLargestComponent}) { | |
| 402 $Molecule->KeepLargestComponent(); | |
| 403 } | |
| 404 | |
| 405 if (!$Molecule->DetectRings()) { | |
| 406 return undef; | |
| 407 } | |
| 408 $Molecule->SetAromaticityModel($OptionsInfo{AromaticityModel}); | |
| 409 $Molecule->DetectAromaticity(); | |
| 410 | |
| 411 if ($OptionsInfo{AddHydrogens}) { | |
| 412 $Molecule->AddHydrogens(); | |
| 413 } | |
| 414 | |
| 415 # Calculate physicochemical properties... | |
| 416 $MolecularDescriptorsGenerator->SetMolecule($Molecule); | |
| 417 $MolecularDescriptorsGenerator->GenerateDescriptors(); | |
| 418 | |
| 419 if (!$MolecularDescriptorsGenerator->IsDescriptorsGenerationSuccessful()) { | |
| 420 return undef; | |
| 421 } | |
| 422 | |
| 423 %CalculatedPhysicochemicalProperties = $MolecularDescriptorsGenerator->GetDescriptorNamesAndValues(); | |
| 424 | |
| 425 # Count RuleOf3 violations... | |
| 426 if ($OptionsInfo{RuleOf3Violations}) { | |
| 427 CalculateRuleViolationsCount('RuleOf3Violations', \%CalculatedPhysicochemicalProperties); | |
| 428 } | |
| 429 | |
| 430 # Count RuleOf5 violations... | |
| 431 if ($OptionsInfo{RuleOf5Violations}) { | |
| 432 CalculateRuleViolationsCount('RuleOf5Violations', \%CalculatedPhysicochemicalProperties); | |
| 433 } | |
| 434 | |
| 435 return \%CalculatedPhysicochemicalProperties; | |
| 436 } | |
| 437 | |
| 438 # Setup molecular descriptor generator to calculate property values for specifed | |
| 439 # property names... | |
| 440 # | |
| 441 sub SetupMolecularDescriptorsGenerator { | |
| 442 my($PropertyName, $MolecularDescriptorsGenerator); | |
| 443 | |
| 444 $MolecularDescriptorsGenerator = new MolecularDescriptors::MolecularDescriptorsGenerator('Mode' => 'Specify', 'DescriptorNames' => \@{$OptionsInfo{SpecifiedPropertyNames}}); | |
| 445 | |
| 446 # Setup molecular desciptor calculation parameters... | |
| 447 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('MolecularWeight')}) || exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('ExactMass')}) ) { | |
| 448 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'WeightAndMassDescriptors', %{$OptionsInfo{PrecisionParametersMap}}); | |
| 449 } | |
| 450 | |
| 451 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('RotatableBonds')})) { | |
| 452 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'RotatableBondsDescriptors', %{$OptionsInfo{RotatableBondsParametersMap}}); | |
| 453 } | |
| 454 | |
| 455 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('HydrogenBondDonors')}) || exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('HydrogenBondAcceptors')}) ) { | |
| 456 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'HydrogenBondsDescriptors', 'HydrogenBondsType' => $OptionsInfo{HydrogenBonds}); | |
| 457 } | |
| 458 | |
| 459 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('TPSA')})) { | |
| 460 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'TPSADescriptors', %{$OptionsInfo{TPSAParametersMap}}); | |
| 461 } | |
| 462 | |
| 463 if (exists($OptionsInfo{SpecifiedPropertyNamesMap}{lc('MolecularComplexity')})) { | |
| 464 $MolecularDescriptorsGenerator->SetDescriptorClassParameters('DescriptorClassName' => 'MolecularComplexityDescriptors', %{$OptionsInfo{MolecularComplexityParametersMap}}); | |
| 465 } | |
| 466 | |
| 467 return $MolecularDescriptorsGenerator; | |
| 468 } | |
| 469 | |
| 470 # Calculate RuleOf3 or RuleOf5 violations count... | |
| 471 # | |
| 472 sub CalculateRuleViolationsCount { | |
| 473 my($RuleViolationsType, $CalculatedPropertiesMapRef) = @_; | |
| 474 my($RuleViolationsCount, $PropertyName); | |
| 475 | |
| 476 $RuleViolationsCount = 0; | |
| 477 | |
| 478 RULEVIOLATIONSTYPE: { | |
| 479 if ($RuleViolationsType =~ /^RuleOf3Violations$/i) { | |
| 480 for $PropertyName (@{$OptionsInfo{RuleOf3PropertyNames}}) { | |
| 481 if ($CalculatedPropertiesMapRef->{$PropertyName} > $OptionsInfo{RuleOf3MaxPropertyValuesMap}{$PropertyName}) { | |
| 482 $RuleViolationsCount++; | |
| 483 } | |
| 484 } | |
| 485 last RULEVIOLATIONSTYPE; | |
| 486 } | |
| 487 | |
| 488 if ($RuleViolationsType =~ /^RuleOf5Violations$/i) { | |
| 489 for $PropertyName (@{$OptionsInfo{RuleOf5PropertyNames}}) { | |
| 490 if ($CalculatedPropertiesMapRef->{$PropertyName} > $OptionsInfo{RuleOf5MaxPropertyValuesMap}{$PropertyName}) { | |
| 491 $RuleViolationsCount++; | |
| 492 } | |
| 493 } | |
| 494 last RULEVIOLATIONSTYPE; | |
| 495 } | |
| 496 | |
| 497 die "Warning: Unknown rule violation type: $RuleViolationsType..."; | |
| 498 } | |
| 499 | |
| 500 # Set rule violation count... | |
| 501 $CalculatedPropertiesMapRef->{$RuleViolationsType} = $RuleViolationsCount; | |
| 502 | |
| 503 } | |
| 504 | |
| 505 # Retrieve information about SD files... | |
| 506 # | |
| 507 sub RetrieveSDFilesInfo { | |
| 508 my($SDFile, $Index, $FileDir, $FileExt, $FileName, $OutFileRoot, $TextOutFileExt, $SDOutFileExt, $NewSDFileName, $NewTextFileName, $CheckDataField, $CollectDataFields, $AllDataFieldsRef, $CommonDataFieldsRef); | |
| 509 | |
| 510 %SDFilesInfo = (); | |
| 511 @{$SDFilesInfo{FileOkay}} = (); | |
| 512 @{$SDFilesInfo{OutFileRoot}} = (); | |
| 513 @{$SDFilesInfo{SDOutFileNames}} = (); | |
| 514 @{$SDFilesInfo{TextOutFileNames}} = (); | |
| 515 @{$SDFilesInfo{AllDataFieldsRef}} = (); | |
| 516 @{$SDFilesInfo{CommonDataFieldsRef}} = (); | |
| 517 | |
| 518 $CheckDataField = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^CompoundID$/i) && ($OptionsInfo{CompoundIDMode} =~ /^DataField$/i)) ? 1 : 0; | |
| 519 $CollectDataFields = ($OptionsInfo{TextOutput} && ($OptionsInfo{DataFieldsMode} =~ /^(All|Common)$/i)) ? 1 : 0; | |
| 520 | |
| 521 FILELIST: for $Index (0 .. $#SDFilesList) { | |
| 522 $SDFile = $SDFilesList[$Index]; | |
| 523 | |
| 524 $SDFilesInfo{FileOkay}[$Index] = 0; | |
| 525 $SDFilesInfo{OutFileRoot}[$Index] = ''; | |
| 526 $SDFilesInfo{SDOutFileNames}[$Index] = ''; | |
| 527 $SDFilesInfo{TextOutFileNames}[$Index] = ''; | |
| 528 | |
| 529 $SDFile = $SDFilesList[$Index]; | |
| 530 if (!(-e $SDFile)) { | |
| 531 warn "Warning: Ignoring file $SDFile: It doesn't exist\n"; | |
| 532 next FILELIST; | |
| 533 } | |
| 534 if (!CheckFileType($SDFile, "sd sdf")) { | |
| 535 warn "Warning: Ignoring file $SDFile: It's not a SD file\n"; | |
| 536 next FILELIST; | |
| 537 } | |
| 538 | |
| 539 if ($CheckDataField) { | |
| 540 # Make sure data field exists in SD file.. | |
| 541 my($CmpdString, $SpecifiedDataField, @CmpdLines, %DataFieldValues); | |
| 542 | |
| 543 @CmpdLines = (); | |
| 544 open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n"; | |
| 545 $CmpdString = ReadCmpdString(\*SDFILE); | |
| 546 close SDFILE; | |
| 547 @CmpdLines = split "\n", $CmpdString; | |
| 548 %DataFieldValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines); | |
| 549 $SpecifiedDataField = $OptionsInfo{CompoundID}; | |
| 550 if (!exists $DataFieldValues{$SpecifiedDataField}) { | |
| 551 warn "Warning: Ignoring file $SDFile: Data field value, $SpecifiedDataField, using \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\" doesn't exist\n"; | |
| 552 next FILELIST; | |
| 553 } | |
| 554 } | |
| 555 | |
| 556 $AllDataFieldsRef = ''; | |
| 557 $CommonDataFieldsRef = ''; | |
| 558 if ($CollectDataFields) { | |
| 559 my($CmpdCount); | |
| 560 open SDFILE, "$SDFile" or die "Error: Couldn't open $SDFile: $! \n"; | |
| 561 ($CmpdCount, $AllDataFieldsRef, $CommonDataFieldsRef) = GetAllAndCommonCmpdDataHeaderLabels(\*SDFILE); | |
| 562 close SDFILE; | |
| 563 } | |
| 564 | |
| 565 # Setup output file names... | |
| 566 $FileDir = ""; $FileName = ""; $FileExt = ""; | |
| 567 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile); | |
| 568 | |
| 569 $TextOutFileExt = "csv"; | |
| 570 if ($Options{outdelim} =~ /^tab$/i) { | |
| 571 $TextOutFileExt = "tsv"; | |
| 572 } | |
| 573 $SDOutFileExt = $FileExt; | |
| 574 | |
| 575 if ($OptionsInfo{OutFileRoot} && (@SDFilesList == 1)) { | |
| 576 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($OptionsInfo{OutFileRoot}); | |
| 577 if ($RootFileName && $RootFileExt) { | |
| 578 $FileName = $RootFileName; | |
| 579 } | |
| 580 else { | |
| 581 $FileName = $OptionsInfo{OutFileRoot}; | |
| 582 } | |
| 583 $OutFileRoot = $FileName; | |
| 584 } | |
| 585 else { | |
| 586 $OutFileRoot = "${FileName}PhysicochemicalProperties"; | |
| 587 } | |
| 588 | |
| 589 $NewSDFileName = "${OutFileRoot}.${SDOutFileExt}"; | |
| 590 $NewTextFileName = "${OutFileRoot}.${TextOutFileExt}"; | |
| 591 | |
| 592 if ($OptionsInfo{SDOutput}) { | |
| 593 if ($SDFile =~ /$NewSDFileName/i) { | |
| 594 warn "Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n"; | |
| 595 print "Specify a different name using \"-r --root\" option or use default name.\n"; | |
| 596 next FILELIST; | |
| 597 } | |
| 598 } | |
| 599 | |
| 600 if (!$OptionsInfo{OverwriteFiles}) { | |
| 601 # Check SD and text outout files... | |
| 602 if ($OptionsInfo{SDOutput}) { | |
| 603 if (-e $NewSDFileName) { | |
| 604 warn "Warning: Ignoring file $SDFile: The file $NewSDFileName already exists\n"; | |
| 605 next FILELIST; | |
| 606 } | |
| 607 } | |
| 608 if ($OptionsInfo{TextOutput}) { | |
| 609 if (-e $NewTextFileName) { | |
| 610 warn "Warning: Ignoring file $SDFile: The file $NewTextFileName already exists\n"; | |
| 611 next FILELIST; | |
| 612 } | |
| 613 } | |
| 614 } | |
| 615 | |
| 616 $SDFilesInfo{FileOkay}[$Index] = 1; | |
| 617 | |
| 618 $SDFilesInfo{OutFileRoot}[$Index] = $OutFileRoot; | |
| 619 $SDFilesInfo{SDOutFileNames}[$Index] = $NewSDFileName; | |
| 620 $SDFilesInfo{TextOutFileNames}[$Index] = $NewTextFileName; | |
| 621 | |
| 622 $SDFilesInfo{AllDataFieldsRef}[$Index] = $AllDataFieldsRef; | |
| 623 $SDFilesInfo{CommonDataFieldsRef}[$Index] = $CommonDataFieldsRef; | |
| 624 } | |
| 625 } | |
| 626 | |
| 627 # Process option values... | |
| 628 sub ProcessOptions { | |
| 629 %OptionsInfo = (); | |
| 630 | |
| 631 $OptionsInfo{AromaticityModel} = $Options{aromaticitymodel}; | |
| 632 | |
| 633 # Process property name related options... | |
| 634 ProcessPropertyNamesOption(); | |
| 635 | |
| 636 # Setup RuleOf3 and RuleOf5 violation calculations... | |
| 637 $OptionsInfo{RuleOf3Violations} = ($Options{ruleof3violations} =~ /^Yes$/i) ? 1 : 0; | |
| 638 $OptionsInfo{RuleOf5Violations} = ($Options{ruleof5violations} =~ /^Yes$/i) ? 1 : 0; | |
| 639 | |
| 640 $OptionsInfo{CompoundIDMode} = $Options{compoundidmode}; | |
| 641 $OptionsInfo{CompoundIDLabel} = $Options{compoundidlabel}; | |
| 642 $OptionsInfo{DataFieldsMode} = $Options{datafieldsmode}; | |
| 643 | |
| 644 my(@SpecifiedDataFields); | |
| 645 @SpecifiedDataFields = (); | |
| 646 | |
| 647 @{$OptionsInfo{SpecifiedDataFields}} = (); | |
| 648 $OptionsInfo{CompoundID} = ''; | |
| 649 | |
| 650 if ($Options{datafieldsmode} =~ /^CompoundID$/i) { | |
| 651 if ($Options{compoundidmode} =~ /^DataField$/i) { | |
| 652 if (!$Options{compoundid}) { | |
| 653 die "Error: You must specify a value for \"--CompoundID\" option in \"DataField\" \"--CompoundIDMode\". \n"; | |
| 654 } | |
| 655 $OptionsInfo{CompoundID} = $Options{compoundid}; | |
| 656 } | |
| 657 elsif ($Options{compoundidmode} =~ /^(LabelPrefix|MolNameOrLabelPrefix)$/i) { | |
| 658 $OptionsInfo{CompoundID} = $Options{compoundid} ? $Options{compoundid} : 'Cmpd'; | |
| 659 } | |
| 660 } | |
| 661 elsif ($Options{datafieldsmode} =~ /^Specify$/i) { | |
| 662 if (!$Options{datafields}) { | |
| 663 die "Error: You must specify a value for \"--DataFields\" option in \"Specify\" \"-d, --DataFieldsMode\". \n"; | |
| 664 } | |
| 665 @SpecifiedDataFields = split /\,/, $Options{datafields}; | |
| 666 push @{$OptionsInfo{SpecifiedDataFields}}, @SpecifiedDataFields; | |
| 667 } | |
| 668 | |
| 669 # Types of hydrogen bonds... | |
| 670 $OptionsInfo{HydrogenBonds} = $Options{hydrogenbonds}; | |
| 671 | |
| 672 # Process precision value parameters... | |
| 673 ProcessPrecisionOption(); | |
| 674 | |
| 675 # Process rotatable bonds parameters... | |
| 676 ProcessRotatableBondsOption(); | |
| 677 | |
| 678 # Process TPSA parameters... | |
| 679 ProcessTPSAOption(); | |
| 680 | |
| 681 # Process molecular complexity parameters... | |
| 682 ProcessMolecularComplexityOption(); | |
| 683 | |
| 684 $OptionsInfo{Filter} = ($Options{filter} =~ /^Yes$/i) ? 1 : 0; | |
| 685 | |
| 686 $OptionsInfo{KeepLargestComponent} = ($Options{keeplargestcomponent} =~ /^Yes$/i) ? 1 : 0; | |
| 687 | |
| 688 $OptionsInfo{Output} = $Options{output}; | |
| 689 $OptionsInfo{SDOutput} = ($Options{output} =~ /^(SD|Both)$/i) ? 1 : 0; | |
| 690 $OptionsInfo{TextOutput} = ($Options{output} =~ /^(Text|Both)$/i) ? 1 : 0; | |
| 691 | |
| 692 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /tab/i ) ? "\t" : (($Options{outdelim} =~ /semicolon/i) ? "\;" : "\,"); | |
| 693 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^Yes$/i) ? 1 : 0; | |
| 694 | |
| 695 $OptionsInfo{OverwriteFiles} = $Options{overwrite} ? 1 : 0; | |
| 696 $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : 0; | |
| 697 } | |
| 698 | |
| 699 # Process property name related options... | |
| 700 # | |
| 701 sub ProcessPropertyNamesOption { | |
| 702 | |
| 703 # Setup supported physicochemical properties... | |
| 704 my($SupportedProperty); | |
| 705 | |
| 706 @{$OptionsInfo{SupportedPropertyNames}} = (); | |
| 707 %{$OptionsInfo{SupportedPropertyNamesMap}} = (); | |
| 708 | |
| 709 @{$OptionsInfo{RuleOf5PropertyNames}} = (); | |
| 710 %{$OptionsInfo{RuleOf5MaxPropertyValuesMap}} = (); | |
| 711 | |
| 712 @{$OptionsInfo{RuleOf3PropertyNames}} = (); | |
| 713 %{$OptionsInfo{RuleOf3MaxPropertyValuesMap}} = (); | |
| 714 | |
| 715 @{$OptionsInfo{DefaultPropertyNames}} = (); | |
| 716 | |
| 717 @{$OptionsInfo{SupportedPropertyNames}} = qw(MolecularWeight ExactMass HeavyAtoms Rings AromaticRings MolecularVolume RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP SMR TPSA Fsp3Carbons Sp3Carbons MolecularComplexity); | |
| 718 | |
| 719 @{$OptionsInfo{RuleOf5PropertyNames}} = qw(MolecularWeight HydrogenBondDonors HydrogenBondAcceptors SLogP); | |
| 720 %{$OptionsInfo{RuleOf5MaxPropertyValuesMap}} = ('MolecularWeight' => 500, 'HydrogenBondDonors' => 5, 'HydrogenBondAcceptors' => 10, 'SLogP' => 5); | |
| 721 | |
| 722 @{$OptionsInfo{RuleOf3PropertyNames}} = qw(MolecularWeight RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP TPSA); | |
| 723 %{$OptionsInfo{RuleOf3MaxPropertyValuesMap}} = ('MolecularWeight' => 300, 'RotatableBonds' => 3, 'HydrogenBondDonors' => 3, 'HydrogenBondAcceptors' => 3, 'SLogP' => 3, 'TPSA' => 60); | |
| 724 | |
| 725 @{$OptionsInfo{DefaultPropertyNames}} = qw(MolecularWeight HeavyAtoms MolecularVolume RotatableBonds HydrogenBondDonors HydrogenBondAcceptors SLogP TPSA); | |
| 726 | |
| 727 for $SupportedProperty (@{$OptionsInfo{SupportedPropertyNames}}) { | |
| 728 $OptionsInfo{SupportedPropertyNamesMap}{lc($SupportedProperty)} = $SupportedProperty; | |
| 729 } | |
| 730 | |
| 731 # Process specified properties.... | |
| 732 my($SpecifiedPropertyName, @SpecifiedPropertyNames, %SpecifiedPropertyNamesMap); | |
| 733 | |
| 734 @SpecifiedPropertyNames = (); | |
| 735 %SpecifiedPropertyNamesMap = (); | |
| 736 | |
| 737 @{$OptionsInfo{SpecifiedPropertyNames}} = (); | |
| 738 %{$OptionsInfo{SpecifiedPropertyNamesMap}} = (); | |
| 739 | |
| 740 if ($Options{mode} =~ /^All$/i) { | |
| 741 @SpecifiedPropertyNames = @{$OptionsInfo{SupportedPropertyNames}}; | |
| 742 } | |
| 743 elsif ($Options{mode} =~ /^RuleOf5$/i) { | |
| 744 @SpecifiedPropertyNames = @{$OptionsInfo{RuleOf5PropertyNames}}; | |
| 745 } | |
| 746 elsif ($Options{mode} =~ /^RuleOf3$/i) { | |
| 747 @SpecifiedPropertyNames = @{$OptionsInfo{RuleOf3PropertyNames}}; | |
| 748 } | |
| 749 elsif (IsEmpty($Options{mode})) { | |
| 750 @SpecifiedPropertyNames = @{$OptionsInfo{DefaultPropertyNames}}; | |
| 751 } | |
| 752 else { | |
| 753 # Comma delimited lisr of specified property names... | |
| 754 my($Mode, $PropertyName, @PropertyNames, @UnsupportedPropertyNames); | |
| 755 | |
| 756 $Mode = $Options{mode}; | |
| 757 $Mode =~ s/ //g; | |
| 758 | |
| 759 @PropertyNames = split ",", $Mode; | |
| 760 @UnsupportedPropertyNames = (); | |
| 761 | |
| 762 for $PropertyName (@PropertyNames) { | |
| 763 if (exists($OptionsInfo{SupportedPropertyNamesMap}{lc($PropertyName)})) { | |
| 764 push @SpecifiedPropertyNames, $PropertyName; | |
| 765 } | |
| 766 else { | |
| 767 push @UnsupportedPropertyNames, $PropertyName; | |
| 768 } | |
| 769 } | |
| 770 if (@UnsupportedPropertyNames) { | |
| 771 if (@UnsupportedPropertyNames > 1) { | |
| 772 warn "Error: The physicochemical property names specified - ", JoinWords(\@UnsupportedPropertyNames, ", ", 0)," - for option \"-m --mode\" are not valid.\n"; | |
| 773 } | |
| 774 else { | |
| 775 warn "Error: The physicochemical property name specified, @UnsupportedPropertyNames , for option \"-m --mode\" is not valid.\n"; | |
| 776 } | |
| 777 die "Allowed values:", JoinWords(\@{$OptionsInfo{SupportedPropertyNames}}, ", ", 0), "\n"; | |
| 778 } | |
| 779 if (!@SpecifiedPropertyNames) { | |
| 780 die "Error: No valid physicochemical property names specified for option \"-m --mode\".\n"; | |
| 781 } | |
| 782 } | |
| 783 | |
| 784 # Set up specified property names map... | |
| 785 PROPERTY: for $SpecifiedPropertyName (@SpecifiedPropertyNames) { | |
| 786 if (exists $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)}) { | |
| 787 warn "Warning: The physicochemical property name, $SpecifiedPropertyName, is specified multiple times as value of option \"-m --mode\" .\n"; | |
| 788 next PROPERTY; | |
| 789 } | |
| 790 # Canonical specified property name... | |
| 791 $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($SpecifiedPropertyName)}; | |
| 792 } | |
| 793 | |
| 794 # Make sure for calculation of RuleOf3Violations, all appropriate property names are specified... | |
| 795 if ($Options{ruleof3violations} =~ /^Yes$/i && $Options{mode} =~ /^RuleOf5$/i) { | |
| 796 die "Error: The value specified, $Options{ruleof3violations}, for \"--RuleOf3Violations\" option in \"RuleOf5\" \"-m --Mode\" is not valid. You must specify RuleOf3 value for \"-m --Mode\" to calculate RuleOf3 violations.\n"; | |
| 797 } | |
| 798 | |
| 799 if ($Options{ruleof3violations} =~ /^Yes$/i) { | |
| 800 my($RuleOf3PropertyName, @MissingRuleOf3Names); | |
| 801 | |
| 802 @MissingRuleOf3Names = (); | |
| 803 PROPERTY: for $RuleOf3PropertyName (@{$OptionsInfo{RuleOf3PropertyNames}}) { | |
| 804 if (exists $SpecifiedPropertyNamesMap{lc($RuleOf3PropertyName)}) { | |
| 805 next PROPERTY; | |
| 806 } | |
| 807 push @MissingRuleOf3Names, $RuleOf3PropertyName; | |
| 808 | |
| 809 # Add property name to specified properties names list and map... | |
| 810 push @SpecifiedPropertyNames, $RuleOf3PropertyName; | |
| 811 $SpecifiedPropertyNamesMap{lc($RuleOf3PropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($RuleOf3PropertyName)}; | |
| 812 } | |
| 813 if (@MissingRuleOf3Names) { | |
| 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"; | |
| 815 } | |
| 816 } | |
| 817 | |
| 818 # Make sure for calculation of RuleOf5Violations, all appropriate property names are specified... | |
| 819 if ($Options{ruleof5violations} =~ /^Yes$/i && $Options{mode} =~ /^RuleOf3$/i) { | |
| 820 die "Error: The value specified, $Options{ruleof5violations}, for \"--RuleOf5Violations\" option in \"RuleOf3\" \"-m --Mode\" is not valid. You must specify RuleOf5 value for \"-m --Mode\" to calculate RuleOf5 violations.\n"; | |
| 821 } | |
| 822 | |
| 823 if ($Options{ruleof5violations} =~ /^Yes$/i) { | |
| 824 my($RuleOf5PropertyName, @MissingRuleOf5Names); | |
| 825 | |
| 826 @MissingRuleOf5Names = (); | |
| 827 PROPERTY: for $RuleOf5PropertyName (@{$OptionsInfo{RuleOf5PropertyNames}}) { | |
| 828 if (exists $SpecifiedPropertyNamesMap{lc($RuleOf5PropertyName)}) { | |
| 829 next PROPERTY; | |
| 830 } | |
| 831 push @MissingRuleOf5Names, $RuleOf5PropertyName; | |
| 832 | |
| 833 # Add property name to specified properties names list and map... | |
| 834 push @SpecifiedPropertyNames, $RuleOf5PropertyName; | |
| 835 $SpecifiedPropertyNamesMap{lc($RuleOf5PropertyName)} = $OptionsInfo{SupportedPropertyNamesMap}{lc($RuleOf5PropertyName)}; | |
| 836 } | |
| 837 if (@MissingRuleOf5Names) { | |
| 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"; | |
| 839 } | |
| 840 } | |
| 841 $OptionsInfo{Mode} = $Options{mode}; | |
| 842 | |
| 843 # Setup canonical specified property names corresponding to supported names in mixed case... | |
| 844 my(@SpecifiedCanonicalPropertyNames); | |
| 845 | |
| 846 @SpecifiedCanonicalPropertyNames = (); | |
| 847 for $SpecifiedPropertyName (@SpecifiedPropertyNames) { | |
| 848 push @SpecifiedCanonicalPropertyNames, $SpecifiedPropertyNamesMap{lc($SpecifiedPropertyName)}; | |
| 849 } | |
| 850 @{$OptionsInfo{SpecifiedPropertyNames}} = @SpecifiedCanonicalPropertyNames; | |
| 851 %{$OptionsInfo{SpecifiedPropertyNamesMap}} = %SpecifiedPropertyNamesMap; | |
| 852 | |
| 853 # Based on specified property names, figure out whether hydrogens need to be added before | |
| 854 # calculation of properties... | |
| 855 # | |
| 856 $OptionsInfo{AddHydrogens} = 0; | |
| 857 if (exists($SpecifiedPropertyNamesMap{lc('MolecularVolume')}) || exists($SpecifiedPropertyNamesMap{lc('SLogP')}) || exists($SpecifiedPropertyNamesMap{lc('SMR')})) { | |
| 858 $OptionsInfo{AddHydrogens} = 1; | |
| 859 } | |
| 860 } | |
| 861 | |
| 862 # Process precision option... | |
| 863 # | |
| 864 sub ProcessPrecisionOption { | |
| 865 my($ParameterName, $ParameterValue, %PrecisionParametersMap, %PrecisionParameterNamesMap); | |
| 866 | |
| 867 %{$OptionsInfo{PrecisionParametersMap}} = (); | |
| 868 | |
| 869 %PrecisionParametersMap = ('WeightPrecision' => 2, 'MassPrecision' => 4); | |
| 870 %PrecisionParameterNamesMap = ('molecularweight' => 'WeightPrecision', 'exactmass' => 'MassPrecision'); | |
| 871 | |
| 872 if ($Options{precision}) { | |
| 873 # Process specified values... | |
| 874 my($Index, $SpecifiedPrecision, @SpecifiedPrecisionValuePairs); | |
| 875 | |
| 876 $SpecifiedPrecision = $Options{precision}; | |
| 877 $SpecifiedPrecision =~ s/ //g; | |
| 878 @SpecifiedPrecisionValuePairs = split ",", $SpecifiedPrecision; | |
| 879 if (@SpecifiedPrecisionValuePairs % 2) { | |
| 880 die "Error: Invalid number of values specified using \"--Precision\" option: It must contain even number of values.\n"; | |
| 881 } | |
| 882 for ($Index = 0; (($Index + 1) < @SpecifiedPrecisionValuePairs); $Index += 2 ) { | |
| 883 $ParameterName = $SpecifiedPrecisionValuePairs[$Index]; | |
| 884 $ParameterValue = $SpecifiedPrecisionValuePairs[$Index + 1]; | |
| 885 if (!exists $PrecisionParameterNamesMap{lc($ParameterName)}) { | |
| 886 die "Error: The precision parameter name specified, $ParameterName, for option \"--Precision\" is not valid.\n"; | |
| 887 } | |
| 888 if (!IsPositiveInteger($ParameterValue)) { | |
| 889 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--Precision\" is not valid. Allowed values: positive integer. \n"; | |
| 890 } | |
| 891 $ParameterName = $PrecisionParameterNamesMap{lc($ParameterName)}; | |
| 892 $PrecisionParametersMap{$ParameterName} = $ParameterValue; | |
| 893 } | |
| 894 } | |
| 895 $OptionsInfo{Precision} = $Options{precision}; | |
| 896 %{$OptionsInfo{PrecisionParametersMap}} = %PrecisionParametersMap; | |
| 897 } | |
| 898 | |
| 899 # Process rotatable bonds option... | |
| 900 sub ProcessRotatableBondsOption { | |
| 901 my($ParameterName, $ParameterValue, %RotatableBondsParametersMap, %RotatableBondsParameterNamesMap); | |
| 902 | |
| 903 %{$OptionsInfo{RotatableBondsParametersMap}} = (); | |
| 904 %RotatableBondsParametersMap = ('IgnoreTerminalBonds' => 1, 'IgnoreBondsToTripleBonds' => 1, 'IgnoreAmideBonds' => 1, 'IgnoreThioamideBonds' => 1, 'IgnoreSulfonamideBonds' => 1); | |
| 905 | |
| 906 for $ParameterName (keys %RotatableBondsParametersMap) { | |
| 907 $RotatableBondsParameterNamesMap{lc($ParameterName)} = $ParameterName; | |
| 908 } | |
| 909 | |
| 910 if ($Options{rotatablebonds}) { | |
| 911 # Process specified values... | |
| 912 my($Index, $SpecifiedRotatableBonds, @SpecifiedRotatableBondsValuePairs); | |
| 913 | |
| 914 $SpecifiedRotatableBonds = $Options{rotatablebonds}; | |
| 915 $SpecifiedRotatableBonds =~ s/ //g; | |
| 916 @SpecifiedRotatableBondsValuePairs = split ",", $SpecifiedRotatableBonds; | |
| 917 if (@SpecifiedRotatableBondsValuePairs % 2) { | |
| 918 die "Error: Invalid number of values specified using \"--RotatableBonds\" option: It must contain even number of values.\n"; | |
| 919 } | |
| 920 for ($Index = 0; (($Index + 1) < @SpecifiedRotatableBondsValuePairs); $Index += 2 ) { | |
| 921 $ParameterName = $SpecifiedRotatableBondsValuePairs[$Index]; | |
| 922 $ParameterValue = $SpecifiedRotatableBondsValuePairs[$Index + 1]; | |
| 923 if (!exists $RotatableBondsParameterNamesMap{lc($ParameterName)}) { | |
| 924 die "Error: The rotatable bonds parameter name specified, $ParameterName, for option \"--RotatableBonds\" is not valid.\n"; | |
| 925 } | |
| 926 if ($ParameterValue !~ /^(Yes|No)$/i) { | |
| 927 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--RotatableBonds\" is not valid. Allowed values: Yes or No. \n"; | |
| 928 } | |
| 929 $ParameterName = $RotatableBondsParameterNamesMap{lc($ParameterName)}; | |
| 930 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0; | |
| 931 $RotatableBondsParametersMap{$ParameterName} = $ParameterValue; | |
| 932 } | |
| 933 } | |
| 934 $OptionsInfo{RotatableBonds} = $Options{rotatablebonds}; | |
| 935 %{$OptionsInfo{RotatableBondsParametersMap}} = %RotatableBondsParametersMap; | |
| 936 } | |
| 937 | |
| 938 # Process TPSA option... | |
| 939 # | |
| 940 sub ProcessTPSAOption { | |
| 941 my($ParameterName, $ParameterValue, %TPSAParametersMap, %TPSAParameterNamesMap); | |
| 942 | |
| 943 %{$OptionsInfo{TPSAParametersMap}} = (); | |
| 944 | |
| 945 %TPSAParametersMap = ('IgnorePhosphorus' => 1, 'IgnoreSulfur' => 1); | |
| 946 for $ParameterName (keys %TPSAParametersMap) { | |
| 947 $TPSAParameterNamesMap{lc($ParameterName)} = $ParameterName; | |
| 948 } | |
| 949 | |
| 950 if ($Options{tpsa}) { | |
| 951 # Process specified values... | |
| 952 my($Index, $SpecifiedTPSA, @SpecifiedTPSAValuePairs); | |
| 953 | |
| 954 $SpecifiedTPSA = $Options{tpsa}; | |
| 955 $SpecifiedTPSA =~ s/ //g; | |
| 956 @SpecifiedTPSAValuePairs = split ",", $SpecifiedTPSA; | |
| 957 if (@SpecifiedTPSAValuePairs % 2) { | |
| 958 die "Error: Invalid number of values specified using \"--TPSA\" option: It must contain even number of values.\n"; | |
| 959 } | |
| 960 for ($Index = 0; (($Index + 1) < @SpecifiedTPSAValuePairs); $Index += 2 ) { | |
| 961 $ParameterName = $SpecifiedTPSAValuePairs[$Index]; | |
| 962 $ParameterValue = $SpecifiedTPSAValuePairs[$Index + 1]; | |
| 963 if (!exists $TPSAParameterNamesMap{lc($ParameterName)}) { | |
| 964 die "Error: The TPSA parameter name specified, $ParameterName, for option \"--TPSA\" is not valid.\n"; | |
| 965 } | |
| 966 if ($ParameterValue !~ /^(Yes|No)$/i) { | |
| 967 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--TPSA\" is not valid. Allowed values: Yes or No. \n"; | |
| 968 } | |
| 969 $ParameterName = $TPSAParameterNamesMap{lc($ParameterName)}; | |
| 970 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0; | |
| 971 $TPSAParametersMap{$ParameterName} = $ParameterValue; | |
| 972 } | |
| 973 } | |
| 974 $OptionsInfo{TPSA} = $Options{tpsa}; | |
| 975 %{$OptionsInfo{TPSAParametersMap}} = %TPSAParametersMap; | |
| 976 } | |
| 977 | |
| 978 # Process molecular complexity parameters... | |
| 979 # | |
| 980 sub ProcessMolecularComplexityOption { | |
| 981 my($MolecularComplexityType, $ParameterName, $ParameterValue, @ParameterNames, @ParameterValues, @AtomIdentifierTypeParameters, %ComplexityParametersMap, %ComplexityParameterNamesMap); | |
| 982 | |
| 983 %{$OptionsInfo{MolecularComplexityParametersMap}} = (); | |
| 984 | |
| 985 %ComplexityParametersMap = ('MolecularComplexityType' => '', 'AtomIdentifierType' => '', | |
| 986 'AtomicInvariantsToUse' => '', 'FunctionalClassesToUse' => '', | |
| 987 'MACCSKeysSize' => '166', 'NeighborhoodRadius' => '2', | |
| 988 'MinPathLength' => '1', 'MaxPathLength' => '8', 'UseBondSymbols' => '1', | |
| 989 'MinDistance' => '1', 'MaxDistance' => '10', 'UseTriangleInequality' => '', | |
| 990 'DistanceBinSize' => '2', 'NormalizationMethodology' => 'None'); | |
| 991 | |
| 992 %ComplexityParameterNamesMap = (); | |
| 993 for $ParameterName (keys %ComplexityParametersMap) { | |
| 994 $ComplexityParameterNamesMap{lc($ParameterName)} = $ParameterName; | |
| 995 } | |
| 996 | |
| 997 if ($Options{molecularcomplexity}) { | |
| 998 # Process specified values... | |
| 999 my($Index, $SpecifiedComplexity, @SpecifiedComplexityValuePairs); | |
| 1000 | |
| 1001 $SpecifiedComplexity = $Options{molecularcomplexity}; | |
| 1002 | |
| 1003 @SpecifiedComplexityValuePairs = split ",", $SpecifiedComplexity; | |
| 1004 if (@SpecifiedComplexityValuePairs % 2) { | |
| 1005 die "Error: Invalid number of values specified using \"--MolecularComplexity\" option: It must contain even number of values.\n"; | |
| 1006 } | |
| 1007 | |
| 1008 for ($Index = 0; (($Index + 1) < @SpecifiedComplexityValuePairs); $Index += 2 ) { | |
| 1009 $ParameterName = $SpecifiedComplexityValuePairs[$Index]; | |
| 1010 $ParameterValue = $SpecifiedComplexityValuePairs[$Index + 1]; | |
| 1011 | |
| 1012 $ParameterName = RemoveLeadingAndTrailingWhiteSpaces($ParameterName); | |
| 1013 $ParameterValue = RemoveLeadingAndTrailingWhiteSpaces($ParameterValue); | |
| 1014 | |
| 1015 if (!exists $ComplexityParameterNamesMap{lc($ParameterName)}) { | |
| 1016 die "Error: The molecular complexity parameter name specified, $ParameterName, for option \"--MolecularComplexity\" is not valid.\n"; | |
| 1017 } | |
| 1018 $ParameterName = $ComplexityParameterNamesMap{lc($ParameterName)}; | |
| 1019 | |
| 1020 if ($ParameterName =~ /^AtomicInvariantsToUse$/i) { | |
| 1021 my($AtomSymbolFound); | |
| 1022 | |
| 1023 $AtomSymbolFound = 0; | |
| 1024 @ParameterValues = split(' ', $ParameterValue); | |
| 1025 for $ParameterValue (@ParameterValues) { | |
| 1026 if (!AtomTypes::AtomicInvariantsAtomTypes::IsAtomicInvariantAvailable($ParameterValue)) { | |
| 1027 die "Error: The atomic invariant specified, $ParameterValue, for AtomicInvariantsToUse in option \"--MolecularComplexity\" is not valid.\n"; | |
| 1028 } | |
| 1029 if ($ParameterValue =~ /^(AS|AtomSymbol)$/i) { | |
| 1030 $AtomSymbolFound = 1; | |
| 1031 } | |
| 1032 } | |
| 1033 if (!$AtomSymbolFound) { | |
| 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"; | |
| 1035 } | |
| 1036 $ParameterValue = JoinWords(\@ParameterValues, ",", 0); | |
| 1037 } | |
| 1038 elsif ($ParameterName =~ /^FunctionalClassesToUse$/i) { | |
| 1039 @ParameterValues = split(' ', $ParameterValue); | |
| 1040 for $ParameterValue (@ParameterValues) { | |
| 1041 if (!AtomTypes::FunctionalClassAtomTypes::IsFunctionalClassAvailable($ParameterValue)) { | |
| 1042 die "Error: The functional class specified, $ParameterValue, for FunctionalClassesToUse in option \"--MolecularComplexity\" is not valid.\n"; | |
| 1043 } | |
| 1044 } | |
| 1045 $ParameterValue = JoinWords(\@ParameterValues, ",", 0); | |
| 1046 } | |
| 1047 else { | |
| 1048 if ($ParameterValue =~ / /) { | |
| 1049 $ParameterValue =~ s/ //g; | |
| 1050 } | |
| 1051 if ($ParameterValue =~ /^(Yes|No)$/i) { | |
| 1052 $ParameterValue = ($ParameterValue =~ /^Yes$/i) ? 1 : 0; | |
| 1053 } | |
| 1054 } | |
| 1055 | |
| 1056 if ($ParameterName =~ /^MolecularComplexityType$/i) { | |
| 1057 if ($ParameterValue !~ /^(AtomTypesFingerprints|ExtendedConnectivityFingerprints|MACCSKeys|PathLengthFingerprints|TopologicalAtomPairsFingerprints|TopologicalAtomTripletsFingerprints|TopologicalAtomTorsionsFingerprints|TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) { | |
| 1058 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: AtomTypesFingerprints, ExtendedConnectivityFingerprints, MACCSKeys, PathLengthFingerprints, TopologicalAtomPairsFingerprints, TopologicalAtomTripletsFingerprints, TopologicalAtomTorsionsFingerprints, TopologicalPharmacophoreAtomPairsFingerprints, or TopologicalPharmacophoreAtomTripletsFingerprints..\n"; | |
| 1059 } | |
| 1060 } | |
| 1061 elsif ($ParameterName =~ /^AtomIdentifierType$/i) { | |
| 1062 if ($ParameterValue !~ /^(AtomicInvariantsAtomTypes|FunctionalClassAtomTypes|DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i) { | |
| 1063 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Supported atom identifier types in current release of MayaChemTools: AtomicInvariantsAtomTypes, FunctionalClassAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes and UFFAtomTypes.\n"; | |
| 1064 } | |
| 1065 } | |
| 1066 elsif ($ParameterName =~ /^(MACCSKeysSize|MinPathLength|MaxPathLength|MinDistance|MaxDistance|DistanceBinSize)$/i) { | |
| 1067 if (!IsPositiveInteger($ParameterValue)) { | |
| 1068 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: positive integer. \n"; | |
| 1069 } | |
| 1070 } | |
| 1071 elsif ($ParameterName =~ /^NeighborhoodRadius$/i) { | |
| 1072 if (!(IsInteger($ParameterValue) && $ParameterValue >=0)) { | |
| 1073 die "Error: The parameter value specified, $ParameterValue, for parameter name, $ParameterName in option \"--MolecularComplexity\" is not valid. Allowed values: 0 or positive integer. \n"; | |
| 1074 } | |
| 1075 } | |
| 1076 elsif ($ParameterName =~ /^NormalizationMethodology$/i) { | |
| 1077 if ($ParameterValue !~ /^(None|ByHeavyAtomsCount|ByPossibleKeysCount)$/i) { | |
| 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"; | |
| 1079 } | |
| 1080 } | |
| 1081 $ComplexityParametersMap{$ParameterName} = $ParameterValue; | |
| 1082 } | |
| 1083 | |
| 1084 if ($ComplexityParametersMap{MACCSKeysSize} !~ /^(166|322)$/i) { | |
| 1085 die "Error: The parameter value specified, $ComplexityParametersMap{MACCSKeysSize}, for parameter name, MACCSKeysSize in option \"--MolecularComplexity\" is not valid. Allowed values: 166 or 322\n"; | |
| 1086 } | |
| 1087 if ($ComplexityParametersMap{MinPathLength} > $ComplexityParametersMap{MaxPathLength}) { | |
| 1088 die "Error: The parameter value specified for MinPathLength, $ComplexityParametersMap{MinPathLength}, must be <= MaxPathLength, $ComplexityParametersMap{MaxPathLength} ...\n"; | |
| 1089 } | |
| 1090 if ($ComplexityParametersMap{MinDistance} > $ComplexityParametersMap{MaxDistance}) { | |
| 1091 die "Error: The parameter value specified for MinDistance, $ComplexityParametersMap{MinDistance}, must be <= MaxDistance, $ComplexityParametersMap{MaxDistance} ...\n"; | |
| 1092 } | |
| 1093 } | |
| 1094 | |
| 1095 # Set default parameter values... | |
| 1096 | |
| 1097 if (IsEmpty($ComplexityParametersMap{MolecularComplexityType})) { | |
| 1098 $ComplexityParametersMap{MolecularComplexityType} = 'MACCSKeys'; | |
| 1099 } | |
| 1100 $MolecularComplexityType = $ComplexityParametersMap{MolecularComplexityType}; | |
| 1101 | |
| 1102 | |
| 1103 if (IsEmpty($ComplexityParametersMap{AtomIdentifierType})) { | |
| 1104 $ComplexityParametersMap{AtomIdentifierType} = ($MolecularComplexityType =~ /^(TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) ? "FunctionalClassAtomTypes" : "AtomicInvariantsAtomTypes"; | |
| 1105 } | |
| 1106 | |
| 1107 if (IsEmpty($ComplexityParametersMap{AtomicInvariantsToUse})) { | |
| 1108 my($AtomicInvariantsToUse); | |
| 1109 | |
| 1110 if ($MolecularComplexityType =~ /^(AtomTypesFingerprints|TopologicalAtomPairsFingerprints|TopologicalAtomTripletsFingerprints|TopologicalAtomTorsionsFingerprints)$/i) { | |
| 1111 $AtomicInvariantsToUse = "AS,X,BO,H,FC"; | |
| 1112 } | |
| 1113 elsif ($MolecularComplexityType =~ /^ExtendedConnectivityFingerprints$/i) { | |
| 1114 $AtomicInvariantsToUse = "AS,X,BO,H,FC,MN"; | |
| 1115 } | |
| 1116 else { | |
| 1117 $AtomicInvariantsToUse = "AS"; | |
| 1118 } | |
| 1119 $ComplexityParametersMap{AtomicInvariantsToUse} = $AtomicInvariantsToUse; | |
| 1120 } | |
| 1121 | |
| 1122 if (IsEmpty($ComplexityParametersMap{FunctionalClassesToUse})) { | |
| 1123 my($FunctionalClassesToUse); | |
| 1124 | |
| 1125 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomPairsFingerprints$/i) { | |
| 1126 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H"; | |
| 1127 } | |
| 1128 elsif ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) { | |
| 1129 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H,Ar"; | |
| 1130 } | |
| 1131 else { | |
| 1132 $FunctionalClassesToUse = "HBD,HBA,PI,NI,H,Ar,Hal"; | |
| 1133 } | |
| 1134 $ComplexityParametersMap{FunctionalClassesToUse} = $FunctionalClassesToUse; | |
| 1135 } | |
| 1136 | |
| 1137 my(@AtomicInvariantsToUse); | |
| 1138 @AtomicInvariantsToUse = split ',', $ComplexityParametersMap{AtomicInvariantsToUse}; | |
| 1139 $ComplexityParametersMap{AtomicInvariantsToUse} = \@AtomicInvariantsToUse; | |
| 1140 | |
| 1141 my(@FunctionalClassesToUse); | |
| 1142 @FunctionalClassesToUse = split ',', $ComplexityParametersMap{FunctionalClassesToUse}; | |
| 1143 $ComplexityParametersMap{FunctionalClassesToUse} = \@FunctionalClassesToUse; | |
| 1144 | |
| 1145 if (IsEmpty($ComplexityParametersMap{UseTriangleInequality})) { | |
| 1146 $ComplexityParametersMap{UseTriangleInequality} = 0; | |
| 1147 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) { | |
| 1148 $ComplexityParametersMap{UseTriangleInequality} = 1; | |
| 1149 } | |
| 1150 } | |
| 1151 | |
| 1152 if ($MolecularComplexityType =~ /^(TopologicalPharmacophoreAtomPairsFingerprints|TopologicalPharmacophoreAtomTripletsFingerprints)$/i) { | |
| 1153 if ($ComplexityParametersMap{AtomIdentifierType} !~ /^FunctionalClassAtomTypes$/i) { | |
| 1154 die "Error: The parameter value specified for AtomIdentifierType, $ComplexityParametersMap{AtomIdentifierType}, in option \"--MolecularComplexity\" is not valid for MolecularComplexityType, $MolecularComplexityType: Allowed value: FunctionalClassAtomTypes...\n"; | |
| 1155 } | |
| 1156 } | |
| 1157 | |
| 1158 # Set up approprate paremeter names for specified molecular complexity... | |
| 1159 | |
| 1160 @ParameterNames = (); | |
| 1161 push @ParameterNames, 'MolecularComplexityType'; | |
| 1162 | |
| 1163 @AtomIdentifierTypeParameters = (); | |
| 1164 push @AtomIdentifierTypeParameters, 'AtomIdentifierType'; | |
| 1165 if ($ComplexityParametersMap{AtomIdentifierType} =~ /^AtomicInvariantsAtomTypes$/i) { | |
| 1166 push @AtomIdentifierTypeParameters, 'AtomicInvariantsToUse'; | |
| 1167 } | |
| 1168 elsif ($ComplexityParametersMap{AtomIdentifierType} =~ /^FunctionalClassAtomTypes$/i) { | |
| 1169 push @AtomIdentifierTypeParameters, 'FunctionalClassesToUse'; | |
| 1170 } | |
| 1171 | |
| 1172 COMPLEXITYTYPE: { | |
| 1173 if ($MolecularComplexityType =~ /^AtomTypesFingerprints$/i) { | |
| 1174 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1175 last COMPLEXITYTYPE; | |
| 1176 } | |
| 1177 if ($MolecularComplexityType =~ /^ExtendedConnectivityFingerprints$/i) { | |
| 1178 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1179 push @ParameterNames, ('NeighborhoodRadius', 'NormalizationMethodology'); | |
| 1180 last COMPLEXITYTYPE; | |
| 1181 } | |
| 1182 if ($MolecularComplexityType =~ /^MACCSKeys$/i) { | |
| 1183 push @ParameterNames, 'MACCSKeysSize'; | |
| 1184 last COMPLEXITYTYPE; | |
| 1185 } | |
| 1186 if ($MolecularComplexityType =~ /^PathLengthFingerprints$/i) { | |
| 1187 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1188 push @ParameterNames, ('MinPathLength', 'MaxPathLength', 'UseBondSymbols'); | |
| 1189 last COMPLEXITYTYPE; | |
| 1190 } | |
| 1191 if ($MolecularComplexityType =~ /^TopologicalAtomPairsFingerprints$/i) { | |
| 1192 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1193 push @ParameterNames, ('MinDistance', 'MaxDistance'); | |
| 1194 last COMPLEXITYTYPE; | |
| 1195 } | |
| 1196 if ($MolecularComplexityType =~ /^TopologicalAtomTripletsFingerprints$/i) { | |
| 1197 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1198 push @ParameterNames, ('MinDistance', 'MaxDistance', 'UseTriangleInequality'); | |
| 1199 last COMPLEXITYTYPE; | |
| 1200 } | |
| 1201 if ($MolecularComplexityType =~ /^TopologicalAtomTorsionsFingerprints$/i) { | |
| 1202 push @ParameterNames, @AtomIdentifierTypeParameters; | |
| 1203 last COMPLEXITYTYPE; | |
| 1204 } | |
| 1205 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomPairsFingerprints$/i) { | |
| 1206 push @ParameterNames, ('AtomIdentifierType', 'FunctionalClassesToUse', 'MinDistance', 'MaxDistance', 'NormalizationMethodology'); | |
| 1207 last COMPLEXITYTYPE; | |
| 1208 } | |
| 1209 if ($MolecularComplexityType =~ /^TopologicalPharmacophoreAtomTripletsFingerprints$/i) { | |
| 1210 push @ParameterNames, ('AtomIdentifierType', 'FunctionalClassesToUse', 'MinDistance', 'MaxDistance', 'UseTriangleInequality', 'NormalizationMethodology', 'DistanceBinSize'); | |
| 1211 last COMPLEXITYTYPE; | |
| 1212 } | |
| 1213 die "Error: The parameter value specified, $ParameterValue, for parameter name MolecularComplexityType using \"--MolecularComplexity\" is not valid.\n"; | |
| 1214 } | |
| 1215 | |
| 1216 $OptionsInfo{MolecularComplexity} = $Options{molecularcomplexity}; | |
| 1217 | |
| 1218 %{$OptionsInfo{MolecularComplexityParametersMap}} = (); | |
| 1219 for $ParameterName (@ParameterNames) { | |
| 1220 $ParameterValue = $ComplexityParametersMap{$ParameterName}; | |
| 1221 $OptionsInfo{MolecularComplexityParametersMap}{$ParameterName} = $ParameterValue; | |
| 1222 } | |
| 1223 } | |
| 1224 | |
| 1225 # Setup script usage and retrieve command line arguments specified using various options... | |
| 1226 sub SetupScriptUsage { | |
| 1227 | |
| 1228 # Retrieve all the options... | |
| 1229 %Options = (); | |
| 1230 | |
| 1231 $Options{aromaticitymodel} = 'MayaChemToolsAromaticityModel'; | |
| 1232 | |
| 1233 $Options{compoundidmode} = 'LabelPrefix'; | |
| 1234 $Options{compoundidlabel} = 'CompoundID'; | |
| 1235 $Options{datafieldsmode} = 'CompoundID'; | |
| 1236 | |
| 1237 $Options{filter} = 'Yes'; | |
| 1238 | |
| 1239 $Options{hydrogenbonds} = 'HBondsType2'; | |
| 1240 | |
| 1241 $Options{keeplargestcomponent} = 'Yes'; | |
| 1242 | |
| 1243 # Default mode values are set later... | |
| 1244 $Options{mode} = ''; | |
| 1245 | |
| 1246 # Default moelcular complexity values are set later... | |
| 1247 $Options{molecularcomplexity} = ''; | |
| 1248 | |
| 1249 # Default precision values are set later... | |
| 1250 $Options{precision} = ''; | |
| 1251 | |
| 1252 $Options{output} = 'text'; | |
| 1253 $Options{outdelim} = 'comma'; | |
| 1254 $Options{quote} = 'yes'; | |
| 1255 | |
| 1256 # Default rotatable bond parameter values are set later... | |
| 1257 $Options{rotatablebonds} = ''; | |
| 1258 | |
| 1259 $Options{ruleof3violations} = 'No'; | |
| 1260 $Options{ruleof5violations} = 'No'; | |
| 1261 | |
| 1262 # Default TPSA paramater values are set later... | |
| 1263 $Options{tpsa} = ''; | |
| 1264 | |
| 1265 if (!GetOptions(\%Options, "aromaticitymodel=s", "compoundid=s", "compoundidlabel=s", "compoundidmode=s", "datafields=s", "datafieldsmode|d=s", "filter|f=s", "help|h", "hydrogenbonds=s", "keeplargestcomponent|k=s", "mode|m=s", "molecularcomplexity=s", "outdelim=s", "output=s", "overwrite|o", "precision=s", "rotatablebonds=s", "ruleof3violations=s", "ruleof5violations=s", "quote|q=s", "root|r=s", "tpsa=s", "workingdir|w=s")) { | |
| 1266 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"; | |
| 1267 } | |
| 1268 if ($Options{workingdir}) { | |
| 1269 if (! -d $Options{workingdir}) { | |
| 1270 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"; | |
| 1271 } | |
| 1272 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n"; | |
| 1273 } | |
| 1274 if (!Molecule::IsSupportedAromaticityModel($Options{aromaticitymodel})) { | |
| 1275 my(@SupportedModels) = Molecule::GetSupportedAromaticityModels(); | |
| 1276 die "Error: The value specified, $Options{aromaticitymodel}, for option \"--AromaticityModel\" is not valid. Supported aromaticity models in current release of MayaChemTools: @SupportedModels\n"; | |
| 1277 } | |
| 1278 if ($Options{compoundidmode} !~ /^(DataField|MolName|LabelPrefix|MolNameOrLabelPrefix)$/i) { | |
| 1279 die "Error: The value specified, $Options{compoundidmode}, for option \"--CompoundIDMode\" is not valid. Allowed values: DataField, MolName, LabelPrefix or MolNameOrLabelPrefix\n"; | |
| 1280 } | |
| 1281 if ($Options{datafieldsmode} !~ /^(All|Common|Specify|CompoundID)$/i) { | |
| 1282 die "Error: The value specified, $Options{datafieldsmode}, for option \"-d, --DataFieldsMode\" is not valid. Allowed values: All, Common, Specify or CompoundID\n"; | |
| 1283 } | |
| 1284 if ($Options{filter} !~ /^(Yes|No)$/i) { | |
| 1285 die "Error: The value specified, $Options{filter}, for option \"-f, --Filter\" is not valid. Allowed values: Yes or No\n"; | |
| 1286 } | |
| 1287 if ($Options{hydrogenbonds} !~ /^(HBondsType1|HydrogenBondsType1|HBondsType2|HydrogenBondsType2)$/i) { | |
| 1288 die "Error: The value specified, $Options{hydrogenbonds}, for option \"--HydrogenBonds\" is not valid. Allowed values: HBondsType1, HydrogenBondsType1, HBondsType2, HydrogenBondsType2\n"; | |
| 1289 } | |
| 1290 if ($Options{keeplargestcomponent} !~ /^(Yes|No)$/i) { | |
| 1291 die "Error: The value specified, $Options{keeplargestcomponent}, for option \"-k, --KeepLargestComponent\" is not valid. Allowed values: Yes or No\n"; | |
| 1292 } | |
| 1293 if ($Options{output} !~ /^(SD|text|both)$/i) { | |
| 1294 die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: SD, text, or both\n"; | |
| 1295 } | |
| 1296 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) { | |
| 1297 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"; | |
| 1298 } | |
| 1299 if ($Options{quote} !~ /^(Yes|No)$/i) { | |
| 1300 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: Yes or No\n"; | |
| 1301 } | |
| 1302 if ($Options{ruleof3violations} !~ /^(Yes|No)$/i) { | |
| 1303 die "Error: The value specified, $Options{ruleof3violations}, for option \"--RuleOf3Violations\" is not valid. Allowed values: Yes or No\n"; | |
| 1304 } | |
| 1305 if ($Options{ruleof5violations} !~ /^(Yes|No)$/i) { | |
| 1306 die "Error: The value specified, $Options{ruleof5violations}, for option \"--RuleOf5Violations\" is not valid. Allowed values: Yes or No\n"; | |
| 1307 } | |
| 1308 } | |
| 1309 | |
| 1310 __END__ | |
| 1311 | |
| 1312 =head1 NAME | |
| 1313 | |
| 1314 CalculatePhysicochemicalProperties.pl - Calculate physicochemical properties for SD files | |
| 1315 | |
| 1316 =head1 SYNOPSIS | |
| 1317 | |
| 1318 CalculatePhysicochemicalProperties.pl SDFile(s)... | |
| 1319 | |
| 1320 PhysicochemicalProperties.pl [B<--AromaticityModel> I<AromaticityModelType>] | |
| 1321 [B<--CompoundID> DataFieldName or LabelPrefixString] | |
| 1322 [B<--CompoundIDLabel> text] [B<--CompoundIDMode>] [B<--DataFields> "FieldLabel1, FieldLabel2,..."] | |
| 1323 [B<-d, --DataFieldsMode> All | Common | Specify | CompoundID] [B<-f, --Filter> Yes | No] [B<-h, --help>] | |
| 1324 [B<--HydrogenBonds> HBondsType1 | HBondsType2] [B<-k, --KeepLargestComponent> Yes | No] | |
| 1325 [B<-m, --mode> All | RuleOf5 | RuleOf3 | "name1, [name2,...]"] | |
| 1326 [B<--MolecularComplexity> I<Name,Value, [Name,Value,...]>] | |
| 1327 [B<--OutDelim> comma | tab | semicolon] [B<--output> SD | text | both] [B<-o, --overwrite>] | |
| 1328 [B<--Precision> Name,Number,[Name,Number,..]] [B<--RotatableBonds> Name,Value, [Name,Value,...]] | |
| 1329 [B<--RuleOf3Violations> Yes | No] [B<--RuleOf5Violations> Yes | No] | |
| 1330 [B<-q, --quote> Yes | No] [B<-r, --root> RootName] | |
| 1331 [B<-w, --WorkingDir> dirname] SDFile(s)... | |
| 1332 | |
| 1333 =head1 DESCRIPTION | |
| 1334 | |
| 1335 Calculate physicochemical properties for I<SDFile(s)> and create appropriate SD or CSV/TSV | |
| 1336 text file(s) containing calculated properties. | |
| 1337 | |
| 1338 The current release of MayaChemTools supports the calculation of these physicochemical | |
| 1339 properties: | |
| 1340 | |
| 1341 MolecularWeight, ExactMass, HeavyAtoms, Rings, AromaticRings, | |
| 1342 van der Waals MolecularVolume [ Ref 93 ], RotatableBonds, | |
| 1343 HydrogenBondDonors, HydrogenBondAcceptors, LogP and | |
| 1344 Molar Refractivity (SLogP and SMR) [ Ref 89 ], Topological Polar | |
| 1345 Surface Area (TPSA) [ Ref 90 ], Fraction of SP3 carbons (Fsp3Carbons) | |
| 1346 and SP3 carbons (Sp3Carbons) [ Ref 115-116, Ref 119 ], | |
| 1347 MolecularComplexity [ Ref 117-119 ] | |
| 1348 | |
| 1349 Multiple SDFile names are separated by spaces. The valid file extensions are I<.sdf> | |
| 1350 and I<.sd>. All other file names are ignored. All the SD files in a current directory | |
| 1351 can be specified either by I<*.sdf> or the current directory name. | |
| 1352 | |
| 1353 The calculation of molecular complexity using I<MolecularComplexityType> parameter | |
| 1354 corresponds to the number of bits-set or unique keys [ Ref 117-119 ] in molecular fingerprints. | |
| 1355 Default value for I<MolecularComplexityType>: I<MACCSKeys> of size 166. The calculation | |
| 1356 of MACCSKeys is relatively expensive and can take rather substantial amount of time. | |
| 1357 | |
| 1358 =head1 OPTIONS | |
| 1359 | |
| 1360 =over 4 | |
| 1361 | |
| 1362 =item B<--AromaticityModel> I<MDLAromaticityModel | TriposAromaticityModel | MMFFAromaticityModel | ChemAxonBasicAromaticityModel | ChemAxonGeneralAromaticityModel | DaylightAromaticityModel | MayaChemToolsAromaticityModel> | |
| 1363 | |
| 1364 Specify aromaticity model to use during detection of aromaticity. Possible values in the current | |
| 1365 release are: I<MDLAromaticityModel, TriposAromaticityModel, MMFFAromaticityModel, | |
| 1366 ChemAxonBasicAromaticityModel, ChemAxonGeneralAromaticityModel, DaylightAromaticityModel | |
| 1367 or MayaChemToolsAromaticityModel>. Default value: I<MayaChemToolsAromaticityModel>. | |
| 1368 | |
| 1369 The supported aromaticity model names along with model specific control parameters | |
| 1370 are defined in B<AromaticityModelsData.csv>, which is distributed with the current release | |
| 1371 and is available under B<lib/data> directory. B<Molecule.pm> module retrieves data from | |
| 1372 this file during class instantiation and makes it available to method B<DetectAromaticity> | |
| 1373 for detecting aromaticity corresponding to a specific model. | |
| 1374 | |
| 1375 =item B<--CompoundID> I<DataFieldName or LabelPrefixString> | |
| 1376 | |
| 1377 This value is B<--CompoundIDMode> specific and indicates how compound ID is generated. | |
| 1378 | |
| 1379 For I<DataField> value of B<--CompoundIDMode> option, it corresponds to datafield label name | |
| 1380 whose value is used as compound ID; otherwise, it's a prefix string used for generating compound | |
| 1381 IDs like LabelPrefixString<Number>. Default value, I<Cmpd>, generates compound IDs which | |
| 1382 look like Cmpd<Number>. | |
| 1383 | |
| 1384 Examples for I<DataField> value of B<--CompoundIDMode>: | |
| 1385 | |
| 1386 MolID | |
| 1387 ExtReg | |
| 1388 | |
| 1389 Examples for I<LabelPrefix> or I<MolNameOrLabelPrefix> value of B<--CompoundIDMode>: | |
| 1390 | |
| 1391 Compound | |
| 1392 | |
| 1393 The value specified above generates compound IDs which correspond to Compound<Number> | |
| 1394 instead of default value of Cmpd<Number>. | |
| 1395 | |
| 1396 =item B<--CompoundIDLabel> I<text> | |
| 1397 | |
| 1398 Specify compound ID column label for CSV/TSV text file(s) used during I<CompoundID> value | |
| 1399 of B<--DataFieldsMode> option. Default value: I<CompoundID>. | |
| 1400 | |
| 1401 =item B<--CompoundIDMode> I<DataField | MolName | LabelPrefix | MolNameOrLabelPrefix> | |
| 1402 | |
| 1403 Specify how to generate compound IDs and write to CSV/TSV text file(s) along with calculated | |
| 1404 physicochemical properties for I<text | both> values of B<--output> option: use a I<SDFile(s)> | |
| 1405 datafield value; use molname line from I<SDFile(s)>; generate a sequential ID with specific prefix; | |
| 1406 use combination of both MolName and LabelPrefix with usage of LabelPrefix values for empty | |
| 1407 molname lines. | |
| 1408 | |
| 1409 Possible values: I<DataField | MolName | LabelPrefix | MolNameOrLabelPrefix>. | |
| 1410 Default value: I<LabelPrefix>. | |
| 1411 | |
| 1412 For I<MolNameAndLabelPrefix> value of B<--CompoundIDMode>, molname line in I<SDFile(s)> takes | |
| 1413 precedence over sequential compound IDs generated using I<LabelPrefix> and only empty molname | |
| 1414 values are replaced with sequential compound IDs. | |
| 1415 | |
| 1416 This is only used for I<CompoundID> value of B<--DataFieldsMode> option. | |
| 1417 | |
| 1418 =item B<--DataFields> I<"FieldLabel1,FieldLabel2,..."> | |
| 1419 | |
| 1420 Comma delimited list of I<SDFiles(s)> data fields to extract and write to CSV/TSV text file(s) along | |
| 1421 with calculated physicochemical properties for I<text | both> values of B<--output> option. | |
| 1422 | |
| 1423 This is only used for I<Specify> value of B<--DataFieldsMode> option. | |
| 1424 | |
| 1425 Examples: | |
| 1426 | |
| 1427 Extreg | |
| 1428 MolID,CompoundName | |
| 1429 | |
| 1430 =item B<-d, --DataFieldsMode> I<All | Common | Specify | CompoundID> | |
| 1431 | |
| 1432 Specify how data fields in I<SDFile(s)> are transferred to output CSV/TSV text file(s) along | |
| 1433 with calculated physicochemical properties for I<text | both> values of B<--output> option: | |
| 1434 transfer all SD data field; transfer SD data files common to all compounds; extract specified | |
| 1435 data fields; generate a compound ID using molname line, a compound prefix, or a combination | |
| 1436 of both. Possible values: I<All | Common | specify | CompoundID>. Default value: I<CompoundID>. | |
| 1437 | |
| 1438 =item B<-f, --Filter> I<Yes | No> | |
| 1439 | |
| 1440 Specify whether to check and filter compound data in SDFile(s). Possible values: I<Yes or No>. | |
| 1441 Default value: I<Yes>. | |
| 1442 | |
| 1443 By default, compound data is checked before calculating physiochemical properties and compounds | |
| 1444 containing atom data corresponding to non-element symbols or no atom data are ignored. | |
| 1445 | |
| 1446 =item B<-h, --help> | |
| 1447 | |
| 1448 Print this help message. | |
| 1449 | |
| 1450 =item B<--HydrogenBonds> I<HBondsType1 | HBondsType2> | |
| 1451 | |
| 1452 Parameters to control calculation of hydrogen bond donors and acceptors. Possible values: | |
| 1453 I<HBondsType1, HydrogenBondsType1, HBondsType2, HydrogenBondsType2>. Default value: | |
| 1454 I<HBondsType2> which corresponds to B<RuleOf5> definition for number of hydrogen bond | |
| 1455 donors and acceptors. | |
| 1456 | |
| 1457 The current release of MayaChemTools supports identification of two types of hydrogen bond | |
| 1458 donor and acceptor atoms with these names: | |
| 1459 | |
| 1460 HBondsType1 or HydrogenBondsType1 | |
| 1461 HBondsType2 or HydrogenBondsType2 | |
| 1462 | |
| 1463 The names of these hydrogen bond types are rather arbitrary. However, their definitions have | |
| 1464 specific meaning and are as follows: | |
| 1465 | |
| 1466 HydrogenBondsType1 [ Ref 60-61, Ref 65-66 ]: | |
| 1467 | |
| 1468 Donor: NH, NH2, OH - Any N and O with available H | |
| 1469 Acceptor: N[!H], O - Any N without available H and any O | |
| 1470 | |
| 1471 HydrogenBondsType2 [ Ref 91 ]: | |
| 1472 | |
| 1473 Donor: NH, NH2, OH - N and O with available H | |
| 1474 Acceptor: N, O - And N and O | |
| 1475 | |
| 1476 =item B<-k, --KeepLargestComponent> I<Yes | No> | |
| 1477 | |
| 1478 Calculate physicochemical properties for only the largest component in molecule. Possible values: | |
| 1479 I<Yes or No>. Default value: I<Yes>. | |
| 1480 | |
| 1481 For molecules containing multiple connected components, physicochemical properties can be | |
| 1482 calculated in two different ways: use all connected components or just the largest connected | |
| 1483 component. By default, all atoms except for the largest connected component are | |
| 1484 deleted before calculation of physicochemical properties. | |
| 1485 | |
| 1486 =item B<-m, --mode> I<All | RuleOf5 | RuleOf3 | "name1, [name2,...]"> | |
| 1487 | |
| 1488 Specify physicochemical properties to calculate for SDFile(s): calculate all available physical | |
| 1489 chemical properties; calculate properties corresponding to Rule of 5; or use a comma delimited | |
| 1490 list of supported physicochemical properties. Possible values: I<All | RuleOf5 | RuleOf3 | | |
| 1491 "name1, [name2,...]">. | |
| 1492 | |
| 1493 Default value: I<MolecularWeight, HeavyAtoms, MolecularVolume, RotatableBonds, HydrogenBondDonors, | |
| 1494 HydrogenBondAcceptors, SLogP, TPSA>. These properties are calculated by default. | |
| 1495 | |
| 1496 I<RuleOf5> [ Ref 91 ] includes these properties: I<MolecularWeight, HydrogenBondDonors, HydrogenBondAcceptors, | |
| 1497 SLogP>. I<RuleOf5> states: MolecularWeight <= 500, HydrogenBondDonors <= 5, HydrogenBondAcceptors <= 10, and | |
| 1498 logP <= 5. | |
| 1499 | |
| 1500 I<RuleOf3> [ Ref 92 ] includes these properties: I<MolecularWeight, RotatableBonds, HydrogenBondDonors, | |
| 1501 HydrogenBondAcceptors, SLogP, TPSA>. I<RuleOf3> states: MolecularWeight <= 300, RotatableBonds <= 3, | |
| 1502 HydrogenBondDonors <= 3, HydrogenBondAcceptors <= 3, logP <= 3, and TPSA <= 60. | |
| 1503 | |
| 1504 I<All> calculates all supported physicochemical properties: I<MolecularWeight, ExactMass, | |
| 1505 HeavyAtoms, Rings, AromaticRings, MolecularVolume, RotatableBonds, HydrogenBondDonors, | |
| 1506 HydrogenBondAcceptors, SLogP, SMR, TPSA, Fsp3Carbons, Sp3Carbons, MolecularComplexity>. | |
| 1507 | |
| 1508 =item B<--MolecularComplexity> I<Name,Value, [Name,Value,...]> | |
| 1509 | |
| 1510 Parameters to control calculation of molecular complexity: it's a comma delimited list of parameter | |
| 1511 name and value pairs. | |
| 1512 | |
| 1513 Possible parameter names: I<MolecularComplexityType, AtomIdentifierType, | |
| 1514 AtomicInvariantsToUse, FunctionalClassesToUse, MACCSKeysSize, NeighborhoodRadius, | |
| 1515 MinPathLength, MaxPathLength, UseBondSymbols, MinDistance, MaxDistance, | |
| 1516 UseTriangleInequality, DistanceBinSize, NormalizationMethodology>. | |
| 1517 | |
| 1518 The valid paramater valuse for each parameter name are described in the following sections. | |
| 1519 | |
| 1520 The current release of MayaChemTools supports calculation of molecular complexity using | |
| 1521 I<MolecularComplexityType> parameter corresponding to the number of bits-set or unique | |
| 1522 keys [ Ref 117-119 ] in molecular fingerprints. The valid values for I<MolecularComplexityType> | |
| 1523 are: | |
| 1524 | |
| 1525 AtomTypesFingerprints | |
| 1526 ExtendedConnectivityFingerprints | |
| 1527 MACCSKeys | |
| 1528 PathLengthFingerprints | |
| 1529 TopologicalAtomPairsFingerprints | |
| 1530 TopologicalAtomTripletsFingerprints | |
| 1531 TopologicalAtomTorsionsFingerprints | |
| 1532 TopologicalPharmacophoreAtomPairsFingerprints | |
| 1533 TopologicalPharmacophoreAtomTripletsFingerprints | |
| 1534 | |
| 1535 Default value for I<MolecularComplexityType>: I<MACCSKeys>. | |
| 1536 | |
| 1537 I<AtomIdentifierType> parameter name correspods to atom types used during generation of | |
| 1538 fingerprints. The valid values for I<AtomIdentifierType> are: I<AtomicInvariantsAtomTypes, | |
| 1539 DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, | |
| 1540 SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes>. I<AtomicInvariantsAtomTypes> | |
| 1541 is not supported for during the following values of I<MolecularComplexityType>: I<MACCSKeys, | |
| 1542 TopologicalPharmacophoreAtomPairsFingerprints, TopologicalPharmacophoreAtomTripletsFingerprints>. | |
| 1543 I<FunctionalClassAtomTypes> is the only valid value for I<AtomIdentifierType> for topological | |
| 1544 pharmacophore fingerprints. | |
| 1545 | |
| 1546 Default value for I<AtomIdentifierType>: I<AtomicInvariantsAtomTypes> | |
| 1547 for all except topological pharmacophore fingerprints where it is I<FunctionalClassAtomTypes>. | |
| 1548 | |
| 1549 I<AtomicInvariantsToUse> parameter name and values are used during I<AtomicInvariantsAtomTypes> | |
| 1550 value of parameter I<AtomIdentifierType>. It's a list of space separated valid atomic invariant atom types. | |
| 1551 | |
| 1552 Possible values for atomic invariants are: I<AS, X, BO, LBO, SB, DB, TB, H, Ar, RA, FC, MN, SM>. | |
| 1553 Default value for I<AtomicInvariantsToUse> parameter are set differently for different fingerprints | |
| 1554 using I<MolecularComplexityType> parameter as shown below: | |
| 1555 | |
| 1556 MolecularComplexityType AtomicInvariantsToUse | |
| 1557 | |
| 1558 AtomTypesFingerprints AS X BO H FC | |
| 1559 TopologicalAtomPairsFingerprints AS X BO H FC | |
| 1560 TopologicalAtomTripletsFingerprints AS X BO H FC | |
| 1561 TopologicalAtomTorsionsFingerprints AS X BO H FC | |
| 1562 | |
| 1563 ExtendedConnectivityFingerprints AS X BO H FC MN | |
| 1564 PathLengthFingerprints AS | |
| 1565 | |
| 1566 | |
| 1567 The atomic invariants abbreviations correspond to: | |
| 1568 | |
| 1569 AS = Atom symbol corresponding to element symbol | |
| 1570 | |
| 1571 X<n> = Number of non-hydrogen atom neighbors or heavy atoms | |
| 1572 BO<n> = Sum of bond orders to non-hydrogen atom neighbors or heavy atoms | |
| 1573 LBO<n> = Largest bond order of non-hydrogen atom neighbors or heavy atoms | |
| 1574 SB<n> = Number of single bonds to non-hydrogen atom neighbors or heavy atoms | |
| 1575 DB<n> = Number of double bonds to non-hydrogen atom neighbors or heavy atoms | |
| 1576 TB<n> = Number of triple bonds to non-hydrogen atom neighbors or heavy atoms | |
| 1577 H<n> = Number of implicit and explicit hydrogens for atom | |
| 1578 Ar = Aromatic annotation indicating whether atom is aromatic | |
| 1579 RA = Ring atom annotation indicating whether atom is a ring | |
| 1580 FC<+n/-n> = Formal charge assigned to atom | |
| 1581 MN<n> = Mass number indicating isotope other than most abundant isotope | |
| 1582 SM<n> = Spin multiplicity of atom. Possible values: 1 (singlet), 2 (doublet) or | |
| 1583 3 (triplet) | |
| 1584 | |
| 1585 Atom type generated by AtomTypes::AtomicInvariantsAtomTypes class corresponds to: | |
| 1586 | |
| 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> | |
| 1588 | |
| 1589 Except for AS which is a required atomic invariant in atom types, all other atomic invariants are | |
| 1590 optional. Atom type specification doesn't include atomic invariants with zero or undefined values. | |
| 1591 | |
| 1592 In addition to usage of abbreviations for specifying atomic invariants, the following descriptive words | |
| 1593 are also allowed: | |
| 1594 | |
| 1595 X : NumOfNonHydrogenAtomNeighbors or NumOfHeavyAtomNeighbors | |
| 1596 BO : SumOfBondOrdersToNonHydrogenAtoms or SumOfBondOrdersToHeavyAtoms | |
| 1597 LBO : LargestBondOrderToNonHydrogenAtoms or LargestBondOrderToHeavyAtoms | |
| 1598 SB : NumOfSingleBondsToNonHydrogenAtoms or NumOfSingleBondsToHeavyAtoms | |
| 1599 DB : NumOfDoubleBondsToNonHydrogenAtoms or NumOfDoubleBondsToHeavyAtoms | |
| 1600 TB : NumOfTripleBondsToNonHydrogenAtoms or NumOfTripleBondsToHeavyAtoms | |
| 1601 H : NumOfImplicitAndExplicitHydrogens | |
| 1602 Ar : Aromatic | |
| 1603 RA : RingAtom | |
| 1604 FC : FormalCharge | |
| 1605 MN : MassNumber | |
| 1606 SM : SpinMultiplicity | |
| 1607 | |
| 1608 I<AtomTypes::AtomicInvariantsAtomTypes> module is used to assign atomic invariant | |
| 1609 atom types. | |
| 1610 | |
| 1611 I<FunctionalClassesToUse> parameter name and values are used during I<FunctionalClassAtomTypes> | |
| 1612 value of parameter I<AtomIdentifierType>. It's a list of space separated valid atomic invariant atom types. | |
| 1613 | |
| 1614 Possible values for atom functional classes are: I<Ar, CA, H, HBA, HBD, Hal, NI, PI, RA>. | |
| 1615 | |
| 1616 Default value for I<FunctionalClassesToUse> parameter is set to: | |
| 1617 | |
| 1618 HBD HBA PI NI Ar Hal | |
| 1619 | |
| 1620 for all fingerprints except for the following two I<MolecularComplexityType> fingerints: | |
| 1621 | |
| 1622 MolecularComplexityType FunctionalClassesToUse | |
| 1623 | |
| 1624 TopologicalPharmacophoreAtomPairsFingerprints HBD HBA P, NI H | |
| 1625 TopologicalPharmacophoreAtomTripletsFingerprints HBD HBA PI NI H Ar | |
| 1626 | |
| 1627 The functional class abbreviations correspond to: | |
| 1628 | |
| 1629 HBD: HydrogenBondDonor | |
| 1630 HBA: HydrogenBondAcceptor | |
| 1631 PI : PositivelyIonizable | |
| 1632 NI : NegativelyIonizable | |
| 1633 Ar : Aromatic | |
| 1634 Hal : Halogen | |
| 1635 H : Hydrophobic | |
| 1636 RA : RingAtom | |
| 1637 CA : ChainAtom | |
| 1638 | |
| 1639 Functional class atom type specification for an atom corresponds to: | |
| 1640 | |
| 1641 Ar.CA.H.HBA.HBD.Hal.NI.PI.RA | |
| 1642 | |
| 1643 I<AtomTypes::FunctionalClassAtomTypes> module is used to assign functional class atom | |
| 1644 types. It uses following definitions [ Ref 60-61, Ref 65-66 ]: | |
| 1645 | |
| 1646 HydrogenBondDonor: NH, NH2, OH | |
| 1647 HydrogenBondAcceptor: N[!H], O | |
| 1648 PositivelyIonizable: +, NH2 | |
| 1649 NegativelyIonizable: -, C(=O)OH, S(=O)OH, P(=O)OH | |
| 1650 | |
| 1651 I<MACCSKeysSize> parameter name is only used during I<MACCSKeys> value of | |
| 1652 I<MolecularComplexityType> and corresponds to the size of MACCS key set. Possible | |
| 1653 values: I<166 or 322>. Default value: I<166>. | |
| 1654 | |
| 1655 I<NeighborhoodRadius> parameter name is only used during I<ExtendedConnectivityFingerprints> | |
| 1656 value of I<MolecularComplexityType> and corresponds to atomic neighborhoods radius for | |
| 1657 generating extended connectivity fingerprints. Possible values: positive integer. Default value: | |
| 1658 I<2>. | |
| 1659 | |
| 1660 I<MinPathLength> and I<MaxPathLength> parameters are only used during I<PathLengthFingerprints> | |
| 1661 value of I<MolecularComplexityType> and correspond to minimum and maximum path lengths to use | |
| 1662 for generating path length fingerprints. Possible values: positive integers. Default value: I<MinPathLength - 1>; | |
| 1663 I<MaxPathLength - 8>. | |
| 1664 | |
| 1665 I<UseBondSymbols> parameter is only used during I<PathLengthFingerprints> value of | |
| 1666 I<MolecularComplexityType> and indicates whether bond symbols are included in atom path | |
| 1667 strings used to generate path length fingerprints. Possible value: I<Yes or No>. Default value: | |
| 1668 I<Yes>. | |
| 1669 | |
| 1670 I<MinDistance> and I<MaxDistance> parameters are only used during I<TopologicalAtomPairsFingerprints> | |
| 1671 and I<TopologicalAtomTripletsFingerprints> values of I<MolecularComplexityType> and correspond to | |
| 1672 minimum and maximum bond distance between atom pairs during topological pharmacophore fingerprints. | |
| 1673 Possible values: positive integers. Default value: I<MinDistance - 1>; I<MaxDistance - 10>. | |
| 1674 | |
| 1675 I<UseTriangleInequality> parameter is used during these values for I<MolecularComplexityType>: | |
| 1676 I<TopologicalAtomTripletsFingerprints> and I<TopologicalPharmacophoreAtomTripletsFingerprints>. | |
| 1677 Possible values: I<Yes or No>. It determines wheter to apply triangle inequality to distance triplets. | |
| 1678 Default value: I<TopologicalAtomTripletsFingerprints - No>; | |
| 1679 I<TopologicalPharmacophoreAtomTripletsFingerprints - Yes>. | |
| 1680 | |
| 1681 I<DistanceBinSize> parameter is used during I<TopologicalPharmacophoreAtomTripletsFingerprints> | |
| 1682 value of I<MolecularComplexityType> and correspons to distance bin size used for binning | |
| 1683 distances during generation of topological pharmacophore atom triplets fingerprints. Possible | |
| 1684 value: positive integer. Default value: I<2>. | |
| 1685 | |
| 1686 I<NormalizationMethodology> is only used for these values for I<MolecularComplexityType>: | |
| 1687 I<ExtendedConnectivityFingerprints>, I<TopologicalPharmacophoreAtomPairsFingerprints> | |
| 1688 and I<TopologicalPharmacophoreAtomTripletsFingerprints>. It corresponds to normalization | |
| 1689 methodology to use for scaling the number of bits-set or unique keys during generation of | |
| 1690 fingerprints. Possible values during I<ExtendedConnectivityFingerprints>: I<None or | |
| 1691 ByHeavyAtomsCount>; Default value: I<None>. Possible values during topological | |
| 1692 pharmacophore atom pairs and tripletes fingerprints: I<None or ByPossibleKeysCount>; | |
| 1693 Default value: I<None>. I<ByPossibleKeysCount> corresponds to total number of | |
| 1694 possible topological pharmacophore atom pairs or triplets in a molecule. | |
| 1695 | |
| 1696 Examples of I<MolecularComplexity> name and value parameters: | |
| 1697 | |
| 1698 MolecularComplexityType,AtomTypesFingerprints,AtomIdentifierType, | |
| 1699 AtomicInvariantsAtomTypes,AtomicInvariantsToUse,AS X BO H FC | |
| 1700 | |
| 1701 MolecularComplexityType,ExtendedConnectivityFingerprints, | |
| 1702 AtomIdentifierType,AtomicInvariantsAtomTypes, | |
| 1703 AtomicInvariantsToUse,AS X BO H FC MN,NeighborhoodRadius,2, | |
| 1704 NormalizationMethodology,None | |
| 1705 | |
| 1706 MolecularComplexityType,MACCSKeys,MACCSKeysSize,166 | |
| 1707 | |
| 1708 MolecularComplexityType,PathLengthFingerprints,AtomIdentifierType, | |
| 1709 AtomicInvariantsAtomTypes,AtomicInvariantsToUse,AS,MinPathLength, | |
| 1710 1,MaxPathLength,8,UseBondSymbols,Yes | |
| 1711 | |
| 1712 MolecularComplexityType,TopologicalAtomPairsFingerprints, | |
| 1713 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse, | |
| 1714 AS X BO H FC,MinDistance,1,MaxDistance,10 | |
| 1715 | |
| 1716 MolecularComplexityType,TopologicalAtomTripletsFingerprints, | |
| 1717 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse, | |
| 1718 AS X BO H FC,MinDistance,1,MaxDistance,10,UseTriangleInequality,No | |
| 1719 | |
| 1720 MolecularComplexityType,TopologicalAtomTorsionsFingerprints, | |
| 1721 AtomIdentifierType,AtomicInvariantsAtomTypes,AtomicInvariantsToUse, | |
| 1722 AS X BO H FC | |
| 1723 | |
| 1724 MolecularComplexityType,TopologicalPharmacophoreAtomPairsFingerprints, | |
| 1725 AtomIdentifierType,FunctionalClassAtomTypes,FunctionalClassesToUse, | |
| 1726 HBD HBA PI NI H,MinDistance,1,MaxDistance,10,NormalizationMethodology, | |
| 1727 None | |
| 1728 | |
| 1729 MolecularComplexityType,TopologicalPharmacophoreAtomTripletsFingerprints, | |
| 1730 AtomIdentifierType,FunctionalClassAtomTypes,FunctionalClassesToUse, | |
| 1731 HBD HBA PI NI H Ar,MinDistance,1,MaxDistance,10,NormalizationMethodology, | |
| 1732 None,UseTriangleInequality,Yes,NormalizationMethodology,None, | |
| 1733 DistanceBinSize,2 | |
| 1734 | |
| 1735 =item B<--OutDelim> I<comma | tab | semicolon> | |
| 1736 | |
| 1737 Delimiter for output CSV/TSV text file(s). Possible values: I<comma, tab, or semicolon> | |
| 1738 Default value: I<comma>. | |
| 1739 | |
| 1740 =item B<--output> I<SD | text | both> | |
| 1741 | |
| 1742 Type of output files to generate. Possible values: I<SD, text, or both>. Default value: I<text>. | |
| 1743 | |
| 1744 =item B<-o, --overwrite> | |
| 1745 | |
| 1746 Overwrite existing files. | |
| 1747 | |
| 1748 =item B<--Precision> I<Name,Number,[Name,Number,..]> | |
| 1749 | |
| 1750 Precision of calculated property values in the output file: it's a comma delimited list of | |
| 1751 property name and precision value pairs. Possible property names: I<MolecularWeight, | |
| 1752 ExactMass>. Possible values: positive intergers. Default value: I<MolecularWeight,2, | |
| 1753 ExactMass,4>. | |
| 1754 | |
| 1755 Examples: | |
| 1756 | |
| 1757 ExactMass,3 | |
| 1758 MolecularWeight,1,ExactMass,2 | |
| 1759 | |
| 1760 =item B<-q, --quote> I<Yes | No> | |
| 1761 | |
| 1762 Put quote around column values in output CSV/TSV text file(s). Possible values: | |
| 1763 I<Yes or No>. Default value: I<Yes>. | |
| 1764 | |
| 1765 =item B<-r, --root> I<RootName> | |
| 1766 | |
| 1767 New file name is generated using the root: <Root>.<Ext>. Default for new file names: | |
| 1768 <SDFileName><PhysicochemicalProperties>.<Ext>. The file type determines <Ext> value. | |
| 1769 The sdf, csv, and tsv <Ext> values are used for SD, comma/semicolon, and tab | |
| 1770 delimited text files, respectively.This option is ignored for multiple input files. | |
| 1771 | |
| 1772 =item B<--RotatableBonds> I<Name,Value, [Name,Value,...]> | |
| 1773 | |
| 1774 Parameters to control calculation of rotatable bonds [ Ref 92 ]: it's a comma delimited list of parameter | |
| 1775 name and value pairs. Possible parameter names: I<IgnoreTerminalBonds, IgnoreBondsToTripleBonds, | |
| 1776 IgnoreAmideBonds, IgnoreThioamideBonds, IgnoreSulfonamideBonds>. Possible parameter values: | |
| 1777 I<Yes or No>. By default, value of all parameters is set to I<Yes>. | |
| 1778 | |
| 1779 =item B<--RuleOf3Violations> I<Yes | No> | |
| 1780 | |
| 1781 Specify whether to calculate B<RuleOf3Violations> for SDFile(s). Possible values: I<Yes or No>. | |
| 1782 Default value: I<No>. | |
| 1783 | |
| 1784 For I<Yes> value of B<RuleOf3Violations>, in addition to calculating total number of B<RuleOf3> violations, | |
| 1785 individual violations for compounds are also written to output files. | |
| 1786 | |
| 1787 B<RuleOf3> [ Ref 92 ] states: MolecularWeight <= 300, RotatableBonds <= 3, HydrogenBondDonors <= 3, | |
| 1788 HydrogenBondAcceptors <= 3, logP <= 3, and TPSA <= 60. | |
| 1789 | |
| 1790 =item B<--RuleOf5Violations> I<Yes | No> | |
| 1791 | |
| 1792 Specify whether to calculate B<RuleOf5Violations> for SDFile(s). Possible values: I<Yes or No>. | |
| 1793 Default value: I<No>. | |
| 1794 | |
| 1795 For I<Yes> value of B<RuleOf5Violations>, in addition to calculating total number of B<RuleOf5> violations, | |
| 1796 individual violations for compounds are also written to output files. | |
| 1797 | |
| 1798 B<RuleOf5> [ Ref 91 ] states: MolecularWeight <= 500, HydrogenBondDonors <= 5, HydrogenBondAcceptors <= 10, | |
| 1799 and logP <= 5. | |
| 1800 | |
| 1801 =item B<--TPSA> I<Name,Value, [Name,Value,...]> | |
| 1802 | |
| 1803 Parameters to control calculation of TPSA: it's a comma delimited list of parameter name and value | |
| 1804 pairs. Possible parameter names: I<IgnorePhosphorus, IgnoreSulfur>. Possible parameter values: | |
| 1805 I<Yes or No>. By default, value of all parameters is set to I<Yes>. | |
| 1806 | |
| 1807 By default, TPSA atom contributions from Phosphorus and Sulfur atoms are not included during | |
| 1808 TPSA calculations. [ Ref 91 ] | |
| 1809 | |
| 1810 =item B<-w, --WorkingDir> I<DirName> | |
| 1811 | |
| 1812 Location of working directory. Default value: current directory. | |
| 1813 | |
| 1814 =back | |
| 1815 | |
| 1816 =head1 EXAMPLES | |
| 1817 | |
| 1818 To calculate default set of physicochemical properties - MolecularWeight, HeavyAtoms, | |
| 1819 MolecularVolume, RotatableBonds, HydrogenBondDonor, HydrogenBondAcceptors, SLogP, | |
| 1820 TPSA - and generate a SamplePhysicochemicalProperties.csv file containing sequential | |
| 1821 compound IDs along with properties data, type: | |
| 1822 | |
| 1823 % CalculatePhysicochemicalProperties.pl -o Sample.sdf | |
| 1824 | |
| 1825 To calculate all available physicochemical properties and generate both SampleAllProperties.csv | |
| 1826 and SampleAllProperties.sdf files containing sequential compound IDs in CSV file along with | |
| 1827 properties data, type: | |
| 1828 | |
| 1829 % CalculatePhysicochemicalProperties.pl -m All --output both | |
| 1830 -r SampleAllProperties -o Sample.sdf | |
| 1831 | |
| 1832 To calculate RuleOf5 physicochemical properties and generate a SampleRuleOf5Properties.csv file | |
| 1833 containing sequential compound IDs along with properties data, type: | |
| 1834 | |
| 1835 % CalculatePhysicochemicalProperties.pl -m RuleOf5 | |
| 1836 -r SampleRuleOf5Properties -o Sample.sdf | |
| 1837 | |
| 1838 To calculate RuleOf5 physicochemical properties along with counting RuleOf5 violations and generate | |
| 1839 a SampleRuleOf5Properties.csv file containing sequential compound IDs along with properties data, type: | |
| 1840 | |
| 1841 % CalculatePhysicochemicalProperties.pl -m RuleOf5 --RuleOf5Violations Yes | |
| 1842 -r SampleRuleOf5Properties -o Sample.sdf | |
| 1843 | |
| 1844 To calculate RuleOf3 physicochemical properties and generate a SampleRuleOf3Properties.csv file | |
| 1845 containing sequential compound IDs along with properties data, type: | |
| 1846 | |
| 1847 % CalculatePhysicochemicalProperties.pl -m RuleOf3 | |
| 1848 -r SampleRuleOf3Properties -o Sample.sdf | |
| 1849 | |
| 1850 To calculate RuleOf3 physicochemical properties along with counting RuleOf3 violations and generate | |
| 1851 a SampleRuleOf3Properties.csv file containing sequential compound IDs along with properties data, type: | |
| 1852 | |
| 1853 % CalculatePhysicochemicalProperties.pl -m RuleOf3 --RuleOf3Violations Yes | |
| 1854 -r SampleRuleOf3Properties -o Sample.sdf | |
| 1855 | |
| 1856 To calculate a specific set of physicochemical properties and generate a SampleProperties.csv file | |
| 1857 containing sequential compound IDs along with properties data, type: | |
| 1858 | |
| 1859 % CalculatePhysicochemicalProperties.pl -m "Rings,AromaticRings" | |
| 1860 -r SampleProperties -o Sample.sdf | |
| 1861 | |
| 1862 To calculate HydrogenBondDonors and HydrogenBondAcceptors using HydrogenBondsType1 definition | |
| 1863 and generate a SampleProperties.csv file containing sequential compound IDs along with properties | |
| 1864 data, type: | |
| 1865 | |
| 1866 % CalculatePhysicochemicalProperties.pl -m "HydrogenBondDonors,HydrogenBondAcceptors" | |
| 1867 --HydrogenBonds HBondsType1 -r SampleProperties -o Sample.sdf | |
| 1868 | |
| 1869 To calculate TPSA using sulfur and phosphorus atoms along with nitrogen and oxygen atoms and | |
| 1870 generate a SampleProperties.csv file containing sequential compound IDs along with properties | |
| 1871 data, type: | |
| 1872 | |
| 1873 % CalculatePhysicochemicalProperties.pl -m "TPSA" --TPSA "IgnorePhosphorus,No, | |
| 1874 IgnoreSulfur,No" -r SampleProperties -o Sample.sdf | |
| 1875 | |
| 1876 To calculate MolecularComplexity using extendend connectivity fingerprints corresponding | |
| 1877 to atom neighborhood radius of 2 with atomic invariant atom types without any scaling and | |
| 1878 generate a SampleProperties.csv file containing sequential compound IDs along with properties | |
| 1879 data, type: | |
| 1880 | |
| 1881 % CalculatePhysicochemicalProperties.pl -m MolecularComplexity --MolecularComplexity | |
| 1882 "MolecularComplexityType,ExtendedConnectivityFingerprints,NeighborhoodRadius,2, | |
| 1883 AtomIdentifierType, AtomicInvariantsAtomTypes, | |
| 1884 AtomicInvariantsToUse,AS X BO H FC MN,NormalizationMethodology,None" | |
| 1885 -r SampleProperties -o Sample.sdf | |
| 1886 | |
| 1887 To calculate RuleOf5 physicochemical properties along with counting RuleOf5 violations and generate | |
| 1888 a SampleRuleOf5Properties.csv file containing compound IDs from molecule name line along with | |
| 1889 properties data, type: | |
| 1890 | |
| 1891 % CalculatePhysicochemicalProperties.pl -m RuleOf5 --RuleOf5Violations Yes | |
| 1892 --DataFieldsMode CompoundID --CompoundIDMode MolName | |
| 1893 -r SampleRuleOf5Properties -o Sample.sdf | |
| 1894 | |
| 1895 To calculate all available physicochemical properties and generate a SampleAllProperties.csv | |
| 1896 file containing compound ID using specified data field along with along with properties data, | |
| 1897 type: | |
| 1898 | |
| 1899 % CalculatePhysicochemicalProperties.pl -m All | |
| 1900 --DataFieldsMode CompoundID --CompoundIDMode DataField --CompoundID Mol_ID | |
| 1901 -r SampleAllProperties -o Sample.sdf | |
| 1902 | |
| 1903 To calculate all available physicochemical properties and generate a SampleAllProperties.csv | |
| 1904 file containing compound ID using combination of molecule name line and an explicit compound | |
| 1905 prefix along with properties data, type: | |
| 1906 | |
| 1907 % CalculatePhysicochemicalProperties.pl -m All | |
| 1908 --DataFieldsMode CompoundID --CompoundIDMode MolnameOrLabelPrefix | |
| 1909 --CompoundID Cmpd --CompoundIDLabel MolID -r SampleAllProperties | |
| 1910 -o Sample.sdf | |
| 1911 | |
| 1912 To calculate all available physicochemical properties and generate a SampleAllProperties.csv | |
| 1913 file containing specific data fields columns along with with properties data, type: | |
| 1914 | |
| 1915 % CalculatePhysicochemicalProperties.pl -m All | |
| 1916 --DataFieldsMode Specify --DataFields Mol_ID -r SampleAllProperties | |
| 1917 -o Sample.sdf | |
| 1918 | |
| 1919 To calculate all available physicochemical properties and generate a SampleAllProperties.csv | |
| 1920 file containing common data fields columns along with with properties data, type: | |
| 1921 | |
| 1922 % CalculatePhysicochemicalProperties.pl -m All | |
| 1923 --DataFieldsMode Common -r SampleAllProperties -o Sample.sdf | |
| 1924 | |
| 1925 To calculate all available physicochemical properties and generate both SampleAllProperties.csv | |
| 1926 and CSV files containing all data fields columns in CSV files along with with properties data, type: | |
| 1927 | |
| 1928 % CalculatePhysicochemicalProperties.pl -m All | |
| 1929 --DataFieldsMode All --output both -r SampleAllProperties | |
| 1930 -o Sample.sdf | |
| 1931 | |
| 1932 =head1 AUTHOR | |
| 1933 | |
| 1934 Manish Sud <msud@san.rr.com> | |
| 1935 | |
| 1936 =head1 SEE ALSO | |
| 1937 | |
| 1938 ExtractFromSDtFiles.pl, ExtractFromTextFiles.pl, InfoSDFiles.pl, InfoTextFiles.pl | |
| 1939 | |
| 1940 =head1 COPYRIGHT | |
| 1941 | |
| 1942 Copyright (C) 2015 Manish Sud. All rights reserved. | |
| 1943 | |
| 1944 This file is part of MayaChemTools. | |
| 1945 | |
| 1946 MayaChemTools is free software; you can redistribute it and/or modify it under | |
| 1947 the terms of the GNU Lesser General Public License as published by the Free | |
| 1948 Software Foundation; either version 3 of the License, or (at your option) | |
| 1949 any later version. | |
| 1950 | |
| 1951 =cut |