1 #!/usr/bin/perl -w
   2 #
   3 # File: InfoAminoAcids.pl
   4 # Author: Manish Sud <msud@san.rr.com>
   5 #
   6 # Copyright (C) 2024 Manish Sud. All rights reserved.
   7 #
   8 # This file is part of MayaChemTools.
   9 #
  10 # MayaChemTools is free software; you can redistribute it and/or modify it under
  11 # the terms of the GNU Lesser General Public License as published by the Free
  12 # Software Foundation; either version 3 of the License, or (at your option) any
  13 # later version.
  14 #
  15 # MayaChemTools is distributed in the hope that it will be useful, but without
  16 # any warranty; without even the implied warranty of merchantability of fitness
  17 # for a particular purpose.  See the GNU Lesser General Public License for more
  18 # details.
  19 #
  20 # You should have received a copy of the GNU Lesser General Public License
  21 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
  22 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
  23 # Boston, MA, 02111-1307, USA.
  24 #
  25 
  26 use strict;
  27 use FindBin; use lib "$FindBin::Bin/../lib";
  28 use Getopt::Long;
  29 use File::Basename;
  30 use Text::ParseWords;
  31 use Benchmark;
  32 use FileUtil;
  33 use TextUtil;
  34 use AminoAcids;
  35 
  36 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
  37 
  38 # Autoflush STDOUT
  39 $| = 1;
  40 
  41 # Starting message...
  42 $ScriptName = basename($0);
  43 print "\n$ScriptName: Starting...\n\n";
  44 $StartTime = new Benchmark;
  45 
  46 # Get the options and setup script...
  47 SetupScriptUsage();
  48 if ($Options{help}) {
  49   die GetUsageFromPod("$FindBin::Bin/$ScriptName");
  50 }
  51 
  52 print "Processing options...\n";
  53 my(%OptionsInfo);
  54 ProcessOptions();
  55 
  56 ListAminoAcidProperties();
  57 print "\n$ScriptName:Done...\n\n";
  58 
  59 $EndTime = new Benchmark;
  60 $TotalTime = timediff ($EndTime, $StartTime);
  61 print "Total time: ", timestr($TotalTime), "\n";
  62 
  63 ###############################################################################
  64 
  65 # List data for an amino acid...
  66 sub ListAminoAcidData {
  67   my($DataLabelRef, $DataValueRef) = @_;
  68   my($Index, $Line, $Value);
  69 
  70   if ($OptionsInfo{AminoAcidRowsOutput}) {
  71     $Line = '';
  72     # Format data...
  73     if ($OptionsInfo{OutQuote} || $Options{outdelim} !~ /^comma$/i) {
  74       $Line = JoinWords($DataValueRef, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
  75     }
  76     else {
  77       # Always quote values containing commas...
  78       $Line = ($DataValueRef->[0] =~ /\,/) ? qq("$DataValueRef->[0]") : $DataValueRef->[0];
  79       for $Index (1 .. $#{$DataValueRef} ) {
  80         $Value = $DataValueRef->[$Index];
  81         if ($Value =~ /\,/) {
  82           $Value = qq("$Value");
  83         }
  84         $Line .= $OptionsInfo{OutDelim} . $Value;
  85       }
  86     }
  87     if ($OptionsInfo{FileOutput}) {
  88       print OUTFILE "$Line\n";
  89     }
  90     else {
  91       print "$Line\n";
  92     }
  93   }
  94   else {
  95     # Format and list data...
  96     $Line = '';
  97     for $Index (0 .. $#{$DataLabelRef} ) {
  98       $Line = $DataLabelRef->[$Index] . ' ' . $DataValueRef->[$Index];
  99       if ($OptionsInfo{FileOutput}) {
 100         print OUTFILE "$Line\n";
 101       }
 102       else {
 103         print "$Line\n";
 104       }
 105     }
 106   }
 107 }
 108 
 109 # List data for an amino acid...
 110 sub ListHeaderRowData {
 111   my($DataLabelRef) = @_;
 112   my($Line);
 113 
 114   # Format data...
 115   $Line = JoinWords($DataLabelRef, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 116   $Line =~ s/\://g;
 117   # List data...
 118   if ($OptionsInfo{FileOutput}) {
 119     print OUTFILE "$Line\n";
 120   }
 121   else {
 122     print "$Line\n";
 123   }
 124 }
 125 
 126 # List properties for amino acids...
 127 sub ListAminoAcidProperties {
 128   my($AminoAcidID, $AminoAcidDataRef, $PropertyName, $PropertyValue, @PropertyLabels, @PropertyValues);
 129 
 130   print "Listing information for amino acid(s)...\n";
 131 
 132   if ($OptionsInfo{FileOutput}) {
 133     print "Generating file $OptionsInfo{OutFileName}...\n";
 134     open OUTFILE, ">$OptionsInfo{OutFileName}" or die "Couldn't open $OptionsInfo{OutFileName}: $!\n";
 135   }
 136 
 137   # Setup property labels...
 138   @PropertyLabels = ();
 139   for $PropertyName (@{$OptionsInfo{SpecifiedProperies}}) {
 140     push @PropertyLabels, ("$PropertyName:");
 141   }
 142 
 143   if ($OptionsInfo{AminoAcidRowsOutput}) {
 144     ListHeaderRowData(\@PropertyLabels);
 145   }
 146 
 147   # Go over specified properties...
 148   for $AminoAcidID (@{$OptionsInfo{SpecifiedAminoAcidIDs}}) {
 149     $AminoAcidDataRef = AminoAcids::GetAminoAcidPropertiesData($AminoAcidID);
 150 
 151     if (!$OptionsInfo{AminoAcidRowsOutput}) {
 152       if ($OptionsInfo{FileOutput}) {
 153         print OUTFILE "\nListing properties for amino acid $AminoAcidID...\n\n";
 154       }
 155       else {
 156         print "\nListing properties for amino acid $AminoAcidID...\n\n";
 157       }
 158     }
 159 
 160     # Collect data..
 161     @PropertyValues = ();
 162     for $PropertyName (@{$OptionsInfo{SpecifiedProperies}}) {
 163       $PropertyValue = $AminoAcidDataRef->{$PropertyName};
 164       if (IsFloat($PropertyValue)) {
 165         $PropertyValue = sprintf("%.$OptionsInfo{Precision}f", $PropertyValue) + 0;
 166       }
 167       push @PropertyValues, $PropertyValue;
 168     }
 169     # List data...
 170     ListAminoAcidData(\@PropertyLabels, \@PropertyValues);
 171   }
 172   if ($OptionsInfo{FileOutput}) {
 173     close OUTFILE;
 174   }
 175   print "\n";
 176 }
 177 
 178 # Get propery names from categories...
 179 sub GetPropertyNamesFromCategories {
 180   my($CategoryName) = @_;
 181   my(@PropertyNames);
 182 
 183   @PropertyNames = ();
 184   if ($CategoryName =~ /^Basic$/i) {
 185     @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula','MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4');
 186   } elsif ($CategoryName =~ /^BasicPlus$/i) {
 187     @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'AcidicBasic', 'PolarNonpolar', 'Charged', 'Aromatic', 'HydrophobicHydophilic', 'IsoelectricPoint', 'pKCOOH', 'pKNH3+', 'ChemicalFormula', 'MolecularWeight', 'ExactMass', 'ChemicalFormulaMinusH2O', 'MolecularWeightMinusH2O(18.01524)', 'ExactMassMinusH2O(18.01056)','LinearStructure', 'LinearStructureAtpH7.4');
 188   } elsif ($CategoryName =~ /^BasicAndHydrophobicity$/i) {
 189     @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula', 'MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4', 'HydrophobicityEisenbergAndOthers', 'HydrophobicityHoppAndWoods', 'HydrophobicityJanin', 'HydrophobicityKyteAndDoolittle', 'HydrophobicityRoseAndOthers', 'HydrophobicityWolfendenAndOthers');
 190   } elsif ($CategoryName =~ /^BasicAndHydrophobicityPlus$/i) {
 191     @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');
 192   }
 193 
 194   return @PropertyNames;
 195 }
 196 
 197 # Process option values...
 198 sub ProcessOptions {
 199   %OptionsInfo = ();
 200 
 201   $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
 202   $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
 203 
 204   $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
 205   $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
 206 
 207   $OptionsInfo{OutputStyle} = $Options{outputstyle};
 208 
 209   $OptionsInfo{AminoAcidRowsOutput} = ($Options{outputstyle} =~ /^AminoAcidRows$/i) ? 1 : 0;
 210   $OptionsInfo{FileOutput} = ($Options{output} =~ /^File$/i) ? 1 : 0;
 211 
 212   $OptionsInfo{Precision} = $Options{precision};
 213 
 214   my($AminoAcidID, @AminoAcidIDs);
 215 
 216   @{$OptionsInfo{SpecifiedAminoAcidIDs}} = ();
 217 
 218   # Set up Amino Acids IDs except for All mode...
 219   @AminoAcidIDs = ();
 220 
 221   if (@ARGV >= 1) {
 222     push @AminoAcidIDs, @ARGV;
 223   }
 224   else {
 225     # Setup mode specified default values...
 226     push @AminoAcidIDs, 'Ala';
 227   }
 228 
 229   # Generate list of amino acids...
 230   if (@ARGV == 1 && $ARGV[0] =~ /^All$/i) {
 231     push @{$OptionsInfo{SpecifiedAminoAcidIDs}}, AminoAcids::GetAminoAcids();
 232   }
 233   else {
 234     ID: for $AminoAcidID (@AminoAcidIDs) {
 235       if (AminoAcids::IsAminoAcid($AminoAcidID)) {
 236         push @{$OptionsInfo{SpecifiedAminoAcidIDs}}, $AminoAcidID;
 237       }
 238       else {
 239         warn "Ignoring amino acid ID, $AminoAcidID, specified using command line parameter option: Unknown amino acid ID...\n";
 240         next ID;
 241       }
 242     }
 243   }
 244   SetupSpecifiedProperties();
 245 
 246   # Setup output file name...
 247   $OptionsInfo{OutFileName} = '';
 248   if ($OptionsInfo{FileOutput}) {
 249     my($OutFileRoot, $OutFileExt);
 250 
 251     $OutFileRoot = '';
 252     $OutFileExt = "csv";
 253     if ($Options{outdelim} =~ /^tab$/i) {
 254       $OutFileExt = "tsv";
 255     }
 256     if ($Options{root}) {
 257       my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
 258       if ($RootFileName && $RootFileExt) {
 259         $OutFileRoot = $RootFileName;
 260       }
 261       else {
 262         $OutFileRoot = $Options{root};
 263       }
 264     }
 265     else {
 266       $OutFileRoot = 'AminoAcidsInfo';
 267     }
 268     $OptionsInfo{OutFileName} = $OutFileRoot . '.' . $OutFileExt;
 269     if (!$Options{overwrite}) {
 270       if (-e $OptionsInfo{OutFileName}) {
 271         die "Error: Output file, $OptionsInfo{OutFileName}, already exists.\nUse \-o --overwrite\ option or specify a different name using \"-r --root\" option.\n";
 272       }
 273     }
 274   }
 275 }
 276 
 277 # Setup properties to list...
 278 sub SetupSpecifiedProperties {
 279 
 280   $OptionsInfo{Properties} = defined $Options{properties} ? $Options{properties} : undef;
 281 
 282   $OptionsInfo{PropertiesMode} = $Options{propertiesmode};
 283   $OptionsInfo{PropertiesListing} = $Options{propertieslisting};
 284 
 285   # Make sure appropriate properties/category names are specified...
 286   @{$OptionsInfo{SpecifiedProperies}} = ();
 287   if ($Options{properties} && ($Options{propertiesmode} =~ /^All$/i) ) {
 288     warn "Warning: Ignoring values specifed by \"-p --properties\" option: Not valid for All value of \"--propertiesmode\" option...\n";
 289   }
 290   if ($Options{propertiesmode} =~ /^All$/i) {
 291     if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
 292       push @{$OptionsInfo{SpecifiedProperies}}, AminoAcids::GetAminoAcidPropertiesNames('Alphabetical');
 293     }
 294     else {
 295       push @{$OptionsInfo{SpecifiedProperies}}, AminoAcids::GetAminoAcidPropertiesNames();
 296     }
 297   }
 298   else {
 299     if ($Options{properties}) {
 300       if ($Options{propertiesmode} =~ /^Categories$/i) {
 301         # Check category name...
 302         if ($Options{properties} !~ /^(Basic|BasicPlus|BasicAndHydrophobicity|BasicAndHydrophobicityPlus)$/i) {
 303           die "Error: The value specified, $Options{properties}, for option \"-p --properties\" in conjunction with \"Categories\" value for option \"--propertiesmode\" is not valid. Allowed values: Basic, BasicPlus, BasicAndHydrophobicity, and BasicAndHydrophobicityPlus\n";
 304         }
 305         # Set propertynames...
 306         push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories($Options{properties});
 307       }
 308       else {
 309         # Check property names..
 310         my($Name, $PropertyName, @Names);
 311         @Names = split /\,/, $Options{properties};
 312         NAME: for $Name (@Names) {
 313           $PropertyName = RemoveLeadingAndTrailingWhiteSpaces($Name);
 314           if (AminoAcids::IsAminoAcidProperty($PropertyName)) {
 315             push @{$OptionsInfo{SpecifiedProperies}}, $PropertyName;
 316           }
 317           else {
 318             warn "Warning: Ignoring value, $Name, specifed by \"-p --properties\" option: Unknown property name...\n";
 319           }
 320         }
 321         if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
 322           # ThreeLetterCode, OneLetterCode and AminoAcid are always listed first...
 323           # NaturalIsotopeData in the end...
 324           my($OneLetterCodePresent, $ThreeLetterCodePresent, $AminoAcidPresent,  @AlphabeticalProperties, %PropertiesMap);
 325           %PropertiesMap = ();
 326           @AlphabeticalProperties = ();
 327           $OneLetterCodePresent = 0; $ThreeLetterCodePresent = 0; $AminoAcidPresent = 0;
 328           NAME: for $Name (@{$OptionsInfo{SpecifiedProperies}}) {
 329             if ($Name =~ /^OneLetterCode$/i) {
 330               $OneLetterCodePresent = 1;
 331               next NAME;
 332             }
 333             if ($Name =~ /^ThreeLetterCode$/i) {
 334               $ThreeLetterCodePresent = 1;
 335               next NAME;
 336             }
 337             if ($Name =~ /^AminoAcid$/i) {
 338               $AminoAcidPresent = 1;
 339               next NAME;
 340             }
 341             $PropertiesMap{$Name} = $Name;
 342           }
 343           # Setup the alphabetical list...
 344           if ($ThreeLetterCodePresent) {
 345             push @AlphabeticalProperties, 'ThreeLetterCode';
 346           }
 347           if ($OneLetterCodePresent) {
 348             push @AlphabeticalProperties, 'OneLetterCode';
 349           }
 350           if ($AminoAcidPresent) {
 351             push @AlphabeticalProperties, 'AminoAcid';
 352           }
 353           for $Name (sort keys %PropertiesMap) {
 354             push @AlphabeticalProperties, $Name;
 355           }
 356           @{$OptionsInfo{SpecifiedProperies}} = ();
 357           push @{$OptionsInfo{SpecifiedProperies}}, @AlphabeticalProperties;
 358         }
 359       }
 360     }
 361     else {
 362       # Set default value...
 363       push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories('Basic');
 364     }
 365   }
 366 }
 367 
 368 # Setup script usage  and retrieve command line arguments specified using various options...
 369 sub SetupScriptUsage {
 370 
 371   # Retrieve all the options...
 372   %Options = ();
 373   $Options{outdelim} = "comma";
 374   $Options{output} = "STDOUT";
 375   $Options{outputstyle} = "AminoAcidBlock";
 376   $Options{precision} = 4;
 377   $Options{propertiesmode} = "Categories";
 378   $Options{propertieslisting} = "ByGroup";
 379   $Options{quote} = "yes";
 380 
 381   if (!GetOptions(\%Options, "help|h", "outdelim=s", "output=s", "outputstyle=s", "overwrite|o", "precision=i", "properties|p=s", "propertieslisting=s", "propertiesmode=s", "quote|q=s", "root|r=s", "workingdir|w=s")) {
 382     die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
 383   }
 384   if ($Options{workingdir}) {
 385     if (! -d $Options{workingdir}) {
 386       die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
 387     }
 388     chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
 389   }
 390   if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
 391     die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
 392   }
 393   if ($Options{output} !~ /^(STDOUT|File)$/i) {
 394     die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: STDOUT or File\n";
 395   }
 396   if ($Options{outputstyle} !~ /^(AminoAcidBlock|AminoAcidRows)$/i) {
 397     die "Error: The value specified, $Options{outputstyle}, for option \"--outputstyle\" is not valid. Allowed values: AminoAcidBlock or AminoAcidRows\n";
 398   }
 399   if (!IsPositiveInteger($Options{precision})) {
 400     die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n";
 401   }
 402   if ($Options{propertiesmode} !~ /^(Categories|Names|All)$/i) {
 403     die "Error: The value specified, $Options{propertiesmode}, for option \"--propertiesmode\" is not valid. Allowed values: Categories, Names, or All\n";
 404   }
 405   if ($Options{propertieslisting} !~ /^(ByGroup|Alphabetical)$/i) {
 406     die "Error: The value specified, $Options{propertieslisting}, for option \"--propertieslisting\" is not valid. Allowed values: ByGroup, or Alphabetical\n";
 407   }
 408   if ($Options{quote} !~ /^(yes|no)$/i) {
 409     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
 410   }
 411 }
 412