1 #!/usr/bin/perl -w 2 # 3 # $RCSfile: InfoAminoAcids.pl,v $ 4 # $Date: 2008/01/30 21:44:46 $ 5 # $Revision: 1.12 $ 6 # 7 # Author: Manish Sud <msud@san.rr.com> 8 # 9 # Copyright (C) 2004-2008 Manish Sud. All rights reserved. 10 # 11 # This file is part of MayaChemTools. 12 # 13 # MayaChemTools is free software; you can redistribute it and/or modify it under 14 # the terms of the GNU Lesser General Public License as published by the Free 15 # Software Foundation; either version 3 of the License, or (at your option) any 16 # later version. 17 # 18 # MayaChemTools is distributed in the hope that it will be useful, but without 19 # any warranty; without even the implied warranty of merchantability of fitness 20 # for a particular purpose. See the GNU Lesser General Public License for more 21 # details. 22 # 23 # You should have received a copy of the GNU Lesser General Public License 24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or 25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330, 26 # Boston, MA, 02111-1307, USA. 27 # 28 29 use 5.006; 30 use strict; 31 use FindBin; use lib "$FindBin::Bin/../lib"; 32 use Getopt::Long; 33 use File::Basename; 34 use Text::ParseWords; 35 use Benchmark; 36 use FileUtil; 37 use TextUtil; 38 use AminoAcids; 39 40 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime); 41 42 # Autoflush STDOUT 43 $| = 1; 44 45 # Starting message... 46 $ScriptName = basename($0); 47 print "\n$ScriptName: Starting...\n\n"; 48 $StartTime = new Benchmark; 49 50 # Get the options and setup script... 51 SetupScriptUsage(); 52 if ($Options{help}) { 53 die GetUsageFromPod("$FindBin::Bin/$ScriptName"); 54 } 55 56 my($OutDelim, $OutQuote, $AminoAcidRowsOutput, $FileOutput, $Precision, $OutFileName, @SpecifiedAminoAcidIDs, @SpecifiedProperies,); 57 ProcessOptions(); 58 59 ListAminoAcidProperties(); 60 print "$ScriptName:Done...\n\n"; 61 62 $EndTime = new Benchmark; 63 $TotalTime = timediff ($EndTime, $StartTime); 64 print "Total time: ", timestr($TotalTime), "\n"; 65 66 ############################################################################### 67 68 # Get propery names from categories... 69 sub GetPropertyNamesFromCategories { 70 my($CategoryName) = @_; 71 my(@PropertyNames); 72 73 @PropertyNames = (); 74 if ($CategoryName =~ /^Basic$/i) { 75 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula','MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4'); 76 } elsif ($CategoryName =~ /^BasicPlus$/i) { 77 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'AcidicBasic', 'PolarNonpolar', 'Charged', 'Aromatic', 'HydrophobicHydophilic', 'IsoelectricPoint', 'pKCOOH', 'pKNH3+', 'ChemicalFormula', 'MolecularWeight', 'ExactMass', 'ChemicalFormulaMinusH2O', 'MolecularWeightMinusH2O(18.01524)', 'ExactMassMinusH2O(18.01056)','LinearStructure', 'LinearStructureAtpH7.4'); 78 } elsif ($CategoryName =~ /^BasicAndHydrophobicity$/i) { 79 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula', 'MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4', 'HydrophobicityEisenbergAndOthers', 'HydrophobicityHoppAndWoods', 'HydrophobicityJanin', 'HydrophobicityKyteAndDoolittle', 'HydrophobicityRoseAndOthers', 'HydrophobicityWolfendenAndOthers'); 80 } elsif ($CategoryName =~ /^BasicAndHydrophobicityPlus$/i) { 81 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula', 'MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4', 'HydrophobicityAbrahamAndLeo', 'HydrophobicityBlack', 'HydrophobicityBullAndBreese', 'HydrophobicityChothia', 'HydrophobicityEisenbergAndOthers', 'HydrophobicityFauchereAndOthers', 'HydrophobicityGuy', 'HydrophobicityHPLCAtpH3.4Cowan', 'HydrophobicityHPLCAtpH7.5Cowan', 'HydrophobicityHPLCParkerAndOthers', 'HydrophobicityHPLCWilsonAndOthers', 'HydrophobicityHoppAndWoods', 'HydrophobicityJanin', 'HydrophobicityKyteAndDoolittle', 'HydrophobicityManavalanAndOthers', 'HydrophobicityMiyazawaAndOthers', 'HydrophobicityOMHSweetAndOthers', 'HydrophobicityRaoAndArgos', 'HydrophobicityRfMobility', 'HydrophobicityRoseAndOthers', 'HydrophobicityRoseman', 'HydrophobicityWellingAndOthers', 'HydrophobicityWolfendenAndOthers'); 82 } 83 84 return @PropertyNames; 85 } 86 87 # List data for an amino acid... 88 sub ListAminoAcidData { 89 my($DataLabelRef, $DataValueRef) = @_; 90 my($Index, $Line, $Value); 91 92 if ($AminoAcidRowsOutput) { 93 $Line = ''; 94 # Format data... 95 if ($OutQuote || $Options{outdelim} !~ /^comma$/i) { 96 $Line = JoinWords($DataValueRef, $OutDelim, $OutQuote); 97 } 98 else { 99 # Always quote values containing commas... 100 $Line = ($DataValueRef->[0] =~ /\,/) ? qq("$DataValueRef->[0]") : $DataValueRef->[0]; 101 for $Index (1 .. $#{$DataValueRef} ) { 102 $Value = $DataValueRef->[$Index]; 103 if ($Value =~ /\,/) { 104 $Value = qq("$Value"); 105 } 106 $Line .= $OutDelim . $Value; 107 } 108 } 109 if ($FileOutput) { 110 print OUTFILE "$Line\n"; 111 } 112 else { 113 print "$Line\n"; 114 } 115 } 116 else { 117 # Format and list data... 118 $Line = ''; 119 for $Index (0 .. $#{$DataLabelRef} ) { 120 $Line = $DataLabelRef->[$Index] . ' ' . $DataValueRef->[$Index]; 121 if ($FileOutput) { 122 print OUTFILE "$Line\n"; 123 } 124 else { 125 print "$Line\n"; 126 } 127 } 128 } 129 } 130 131 # List data for an amino acid... 132 sub ListHeaderRowData { 133 my($DataLabelRef) = @_; 134 my($Line); 135 136 # Format data... 137 $Line = JoinWords($DataLabelRef, $OutDelim, $OutQuote); 138 $Line =~ s/\://g; 139 # List data... 140 if ($FileOutput) { 141 print OUTFILE "$Line\n"; 142 } 143 else { 144 print "$Line\n"; 145 } 146 } 147 148 # List properties for amino acids... 149 sub ListAminoAcidProperties { 150 my($AminoAcidID, $AminoAcidDataRef, $PropertyName, $PropertyValue, @PropertyLabels, @PropertyValues); 151 152 print "Listing information for amino acids(s)...\n"; 153 154 if ($FileOutput) { 155 print "Generating file $OutFileName...\n"; 156 open OUTFILE, ">$OutFileName" or die "Couldn't open $OutFileName: $!\n"; 157 } 158 159 # Setup property labels... 160 @PropertyLabels = (); 161 for $PropertyName (@SpecifiedProperies) { 162 push @PropertyLabels, ("$PropertyName:"); 163 } 164 165 if ($AminoAcidRowsOutput) { 166 ListHeaderRowData(\@PropertyLabels); 167 } 168 169 # Go over specified properties... 170 for $AminoAcidID (@SpecifiedAminoAcidIDs) { 171 $AminoAcidDataRef = AminoAcids::GetAminoAcidPropertiesData($AminoAcidID); 172 173 if (!$AminoAcidRowsOutput) { 174 if ($FileOutput) { 175 print OUTFILE "\nListing properties for amino acid $AminoAcidID...\n\n"; 176 } 177 else { 178 print "\nListing properties for amino acid $AminoAcidID...\n\n"; 179 } 180 } 181 182 # Collect data.. 183 @PropertyValues = (); 184 for $PropertyName (@SpecifiedProperies) { 185 $PropertyValue = $AminoAcidDataRef->{$PropertyName}; 186 if (IsFloat($PropertyValue)) { 187 $PropertyValue = sprintf("%.${Precision}f", $PropertyValue) + 0; 188 } 189 push @PropertyValues, $PropertyValue; 190 } 191 # List data... 192 ListAminoAcidData(\@PropertyLabels, \@PropertyValues); 193 } 194 if ($FileOutput) { 195 close OUTFILE; 196 } 197 print "\n"; 198 } 199 200 # Process option values... 201 sub ProcessOptions { 202 $OutDelim = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,"); 203 $OutQuote = ($Options{quote} =~ /^yes$/i) ? 1 : 0; 204 205 $AminoAcidRowsOutput = ($Options{outputstyle} =~ /^AminoAcidRows$/i) ? 1 : 0; 206 $FileOutput = ($Options{output} =~ /^File$/i) ? 1 : 0; 207 208 $Precision = $Options{precision}; 209 210 my($AminoAcidID, @AminoAcidIDs); 211 212 @SpecifiedAminoAcidIDs = (); 213 214 # Set up Amino Acids IDs except for All mode... 215 @AminoAcidIDs = (); 216 217 if (@ARGV >= 1) { 218 push @AminoAcidIDs, @ARGV; 219 } 220 else { 221 # Setup mode specified default values... 222 push @AminoAcidIDs, 'Ala'; 223 } 224 225 # Generate list of amino acids... 226 if (@ARGV == 1 && $ARGV[0] =~ /^All$/i) { 227 push @SpecifiedAminoAcidIDs, AminoAcids::GetAminoAcids(); 228 } 229 else { 230 ID: for $AminoAcidID (@AminoAcidIDs) { 231 if (AminoAcids::IsAminoAcid($AminoAcidID)) { 232 push @SpecifiedAminoAcidIDs, $AminoAcidID; 233 } 234 else { 235 warn "Ignoring amino acid ID, $AminoAcidID, specified using command line parameter option: Unknown amino acid ID...\n"; 236 next ID; 237 } 238 } 239 } 240 SetupSpecifiedProperties(); 241 242 # Setup output file name... 243 $OutFileName = ''; 244 if ($FileOutput) { 245 my($OutFileRoot, $OutFileExt); 246 247 $OutFileRoot = ''; 248 $OutFileExt = "csv"; 249 if ($Options{outdelim} =~ /^tab$/i) { 250 $OutFileExt = "tsv"; 251 } 252 if ($Options{root}) { 253 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root}); 254 if ($RootFileName && $RootFileExt) { 255 $OutFileRoot = $RootFileName; 256 } 257 else { 258 $OutFileRoot = $Options{root}; 259 } 260 } 261 else { 262 $OutFileRoot = 'AminoAcidsInfo'; 263 } 264 $OutFileName = $OutFileRoot . '.' . $OutFileExt; 265 if (!$Options{overwrite}) { 266 if (-e $OutFileName) { 267 die "Error: Output file, $OutFileName, already exists.\nUse \-o --overwrite\ option or specify a different name using \"-r --root\" option.\n"; 268 } 269 } 270 } 271 } 272 273 # Setup properties to list... 274 sub SetupSpecifiedProperties { 275 # Make sure appropriate properties/category names are specified... 276 @SpecifiedProperies = (); 277 if ($Options{properties} && ($Options{propertiesmode} =~ /^All$/i) ) { 278 warn "Warning: Ignoring values specifed by \"-p --properties\" option: Not valid for All value of \"--propertiesmode\" option...\n"; 279 } 280 if ($Options{propertiesmode} =~ /^All$/i) { 281 if ($Options{propertieslisting} =~ /^Alphabetical$/i) { 282 push @SpecifiedProperies, AminoAcids::GetAminoAcidPropertiesNames('Alphabetical'); 283 } 284 else { 285 push @SpecifiedProperies, AminoAcids::GetAminoAcidPropertiesNames(); 286 } 287 } 288 else { 289 if ($Options{properties}) { 290 if ($Options{propertiesmode} =~ /^Categories$/i) { 291 # Check category name... 292 if ($Options{properties} !~ /^(Basic|BasicPlus|BasicAndHydrophobicity|BasicAndHydrophobicityPlus)$/i) { 293 die "Error: The value specified, $Options{properties}, for option \"-p --properties\" in conjunction with \"Categories\" value for option \"--propertiesmode\" is not valid. Allowed values: Basic, BasicPlus, BasicAndHydrophobicity, and BasicAndHydrophobicityPlus\n"; 294 } 295 # Set propertynames... 296 push @SpecifiedProperies, GetPropertyNamesFromCategories($Options{properties}); 297 } 298 else { 299 # Check property names.. 300 my($Name, $PropertyName, @Names); 301 @Names = split /\,/, $Options{properties}; 302 NAME: for $Name (@Names) { 303 $PropertyName = RemoveLeadingAndTrailingWhiteSpaces($Name); 304 if (AminoAcids::IsAminoAcidProperty($PropertyName)) { 305 push @SpecifiedProperies, $PropertyName; 306 } 307 else { 308 warn "Warning: Ignoring value, $Name, specifed by \"-p --properties\" option: Unknown property name...\n"; 309 } 310 } 311 if ($Options{propertieslisting} =~ /^Alphabetical$/i) { 312 # ThreeLetterCode, OneLetterCode and AminoAcid are always listed first... 313 # NaturalIsotopeData in the end... 314 my($OneLetterCodePresent, $ThreeLetterCodePresent, $AminoAcidPresent, @AlphabeticalProperties, %PropertiesMap); 315 %PropertiesMap = (); 316 @AlphabeticalProperties = (); 317 $OneLetterCodePresent = 0; $ThreeLetterCodePresent = 0; $AminoAcidPresent = 0; 318 NAME: for $Name (@SpecifiedProperies) { 319 if ($Name =~ /^OneLetterCode$/i) { 320 $OneLetterCodePresent = 1; 321 next NAME; 322 } 323 if ($Name =~ /^ThreeLetterCode$/i) { 324 $ThreeLetterCodePresent = 1; 325 next NAME; 326 } 327 if ($Name =~ /^AminoAcid$/i) { 328 $AminoAcidPresent = 1; 329 next NAME; 330 } 331 $PropertiesMap{$Name} = $Name; 332 } 333 # Setup the alphabetical list... 334 if ($ThreeLetterCodePresent) { 335 push @AlphabeticalProperties, 'ThreeLetterCode'; 336 } 337 if ($OneLetterCodePresent) { 338 push @AlphabeticalProperties, 'OneLetterCode'; 339 } 340 if ($AminoAcidPresent) { 341 push @AlphabeticalProperties, 'AminoAcid'; 342 } 343 for $Name (sort keys %PropertiesMap) { 344 push @AlphabeticalProperties, $Name; 345 } 346 @SpecifiedProperies = (); 347 push @SpecifiedProperies, @AlphabeticalProperties; 348 } 349 } 350 } 351 else { 352 # Set default value... 353 push @SpecifiedProperies, GetPropertyNamesFromCategories('Basic'); 354 } 355 } 356 } 357 358 # Setup script usage and retrieve command line arguments specified using various options... 359 sub SetupScriptUsage { 360 361 # Retrieve all the options... 362 %Options = (); 363 $Options{outdelim} = "comma"; 364 $Options{output} = "STDOUT"; 365 $Options{outputstyle} = "AminoAcidBlock"; 366 $Options{precision} = 4; 367 $Options{propertiesmode} = "Categories"; 368 $Options{propertieslisting} = "ByGroup"; 369 $Options{quote} = "yes"; 370 371 if (!GetOptions(\%Options, "help|h", "outdelim=s", "output=s", "outputstyle=s", "overwrite|o", "precision=i", "properties|p=s", "propertieslisting=s", "propertiesmode=s", "quote|q=s", "root|r=s", "workingdir|w=s")) { 372 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"; 373 } 374 if ($Options{workingdir}) { 375 if (! -d $Options{workingdir}) { 376 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"; 377 } 378 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n"; 379 } 380 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) { 381 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"; 382 } 383 if ($Options{output} !~ /^(STDOUT|File)$/i) { 384 die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: STDOUT or File\n"; 385 } 386 if ($Options{outputstyle} !~ /^(AminoAcidBlock|AminoAcidRows)$/i) { 387 die "Error: The value specified, $Options{outputstyle}, for option \"--outputstyle\" is not valid. Allowed values: AminoAcidBlock or AminoAcidRows\n"; 388 } 389 if (!IsPositiveInteger($Options{precision})) { 390 die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n"; 391 } 392 if ($Options{propertiesmode} !~ /^(Categories|Names|All)$/i) { 393 die "Error: The value specified, $Options{propertiesmode}, for option \"--propertiesmode\" is not valid. Allowed values: Categories, Names, or All\n"; 394 } 395 if ($Options{propertieslisting} !~ /^(ByGroup|Alphabetical)$/i) { 396 die "Error: The value specified, $Options{propertieslisting}, for option \"--propertieslisting\" is not valid. Allowed values: ByGroup, or Alphabetical\n"; 397 } 398 if ($Options{quote} !~ /^(yes|no)$/i) { 399 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"; 400 } 401 } 402
