forester/java/src/org/forester/application/msa_compactor.java

   1
   2 package org.forester.application;
   3
   4 import java.io.File;
   5 import java.io.FileInputStream;
   6 import java.util.ArrayList;
   7 import java.util.List;
   8
   9 import org.forester.io.parsers.FastaParser;
  10 import org.forester.io.parsers.GeneralMsaParser;
  11 import org.forester.msa.Msa;
  12 import org.forester.msa.Msa.MSA_FORMAT;
  13 import org.forester.msa.MsaMethods;
  14 import org.forester.msa_compactor.MsaCompactor;
  15 import org.forester.util.CommandLineArguments;
  16 import org.forester.util.ForesterUtil;
  17
  18 public class msa_compactor {
  19
  20     final static private String HELP_OPTION_1                          = "help";
  21     final static private String HELP_OPTION_2                          = "h";
  22     final static private String REMOVE_WORST_OFFENDERS_OPTION          = "w";
  23     final static private String AV_GAPINESS_OPTION                     = "a";
  24     final static private String STEP_OPTION                            = "s";
  25     final static private String LENGTH_OPTION                          = "l";
  26     final static private String REALIGN_OPTION                         = "r";
  27     final static private String DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION = "nn";
  28     final static private String PRG_NAME                               = "msa_compactor";
  29     final static private String PRG_DESC                               = "multiple sequnce aligment compactor";
  30     final static private String PRG_VERSION                            = "0.01";
  31     final static private String PRG_DATE                               = "140314";
  32     final static private String E_MAIL                                 = "phylosoft@gmail.com";
  33     final static private String WWW                                    = "https://sites.google.com/site/cmzmasek/home/software/forester";
  34
  35     public static void main( final String args[] ) {
  36         try {
  37             final CommandLineArguments cla = new CommandLineArguments( args );
  38             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
  39                 printHelp();
  40                 System.exit( 0 );
  41             }
  42             final File in = cla.getFile( 0 );
  43             final File out = cla.getFile( 1 );
  44             int worst_remove = -1;
  45             double av = -1;
  46             int length = -1;
  47             int step = 1;
  48             boolean realign = false;
  49             boolean norm = true;
  50             final List<String> allowed_options = new ArrayList<String>();
  51             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
  52             allowed_options.add( AV_GAPINESS_OPTION );
  53             allowed_options.add( LENGTH_OPTION );
  54             allowed_options.add( REALIGN_OPTION );
  55             allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
  56             allowed_options.add( STEP_OPTION );
  57             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  58             if ( dissallowed_options.length() > 0 ) {
  59                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  60             }
  61             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  62                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  63             }
  64             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  65                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  66             }
  67             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  68                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  69             }
  70             if ( cla.isOptionSet( STEP_OPTION ) ) {
  71                 step = cla.getOptionValueAsInt( STEP_OPTION );
  72             }
  73             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  74                 realign = true;
  75             }
  76             if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
  77                 norm = false;
  78             }
  79             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  80             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  81             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  82             //            }
  83             //            else {
  84             //                printHelp();
  85             //                System.exit( 0 );
  86             //            }
  87             Msa msa = null;
  88             final FileInputStream is = new FileInputStream( in );
  89             if ( FastaParser.isLikelyFasta( in ) ) {
  90                 msa = FastaParser.parseMsa( is );
  91             }
  92             else {
  93                 msa = GeneralMsaParser.parse( is );
  94             }
  95             MsaCompactor mc = null;
  96             if ( worst_remove > 0 ) {
  97                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign, norm );
  98             }
  99             else if ( av > 0 ) {
 100                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50 );
 101             }
 102             else if ( length > 0 ) {
 103                 mc = MsaCompactor.reduceLength( msa, length, step, realign );
 104             }
 105             System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
 106             for( final String id : mc.getRemovedSeqIds() ) {
 107                 System.out.println( id );
 108             }
 109             mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
 110         }
 111         catch ( final Exception e ) {
 112             e.printStackTrace();
 113             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 114         }
 115     }
 116
 117     private static void printHelp() {
 118         ForesterUtil.printProgramInformation( PRG_NAME,
 119                                               PRG_DESC,
 120                                               PRG_VERSION,
 121                                               PRG_DATE,
 122                                               E_MAIL,
 123                                               WWW,
 124                                               ForesterUtil.getForesterLibraryInformation() );
 125         System.out.println( "Usage:" );
 126         System.out.println();
 127         System.out.println( PRG_NAME + " <options> <msa input file>" );
 128         System.out.println();
 129         System.out.println( " options: " );
 130         System.out.println();
 131         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>  number of sequences to remove" );
 132         System.out.println( "   -" + REALIGN_OPTION + " to realign using " );
 133         System.out.println();
 134         System.out.println();
 135         System.out.println();
 136     }
 137 }