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.MsaCompactor;
  14 import org.forester.msa.MsaMethods;
  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 PRG_NAME                      = "msa_compactor";
  28     final static private String PRG_DESC                      = "multiple sequnce aligment compactor";
  29     final static private String PRG_VERSION                   = "0.01";
  30     final static private String PRG_DATE                      = "140221";
  31     final static private String E_MAIL                        = "phylosoft@gmail.com";
  32     final static private String WWW                           = "https://sites.google.com/site/cmzmasek/home/software/forester";
  33
  34     public static void main( final String args[] ) {
  35         try {
  36             final CommandLineArguments cla = new CommandLineArguments( args );
  37             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
  38                 printHelp();
  39                 System.exit( 0 );
  40             }
  41             final File in = cla.getFile( 0 );
  42             final File out = cla.getFile( 1 );
  43             int worst_remove = -1;
  44             double av = -1;
  45             int length = -1;
  46             int step = 1;
  47             boolean realign = false;
  48             final List<String> allowed_options = new ArrayList<String>();
  49             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
  50             allowed_options.add( AV_GAPINESS_OPTION );
  51             allowed_options.add( LENGTH_OPTION );
  52             allowed_options.add( REALIGN_OPTION );
  53             allowed_options.add( STEP_OPTION );
  54             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  55             if ( dissallowed_options.length() > 0 ) {
  56                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  57             }
  58             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  59                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  60             }
  61             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  62                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  63             }
  64             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  65                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  66             }
  67             if ( cla.isOptionSet( STEP_OPTION ) ) {
  68                 step = cla.getOptionValueAsInt( STEP_OPTION );
  69             }
  70             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  71                 realign = true;
  72             }
  73             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  74             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  75             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  76             //            }
  77             //            else {
  78             //                printHelp();
  79             //                System.exit( 0 );
  80             //            }
  81             Msa msa = null;
  82             final FileInputStream is = new FileInputStream( in );
  83             if ( FastaParser.isLikelyFasta( in ) ) {
  84                 msa = FastaParser.parseMsa( is );
  85             }
  86             else {
  87                 msa = GeneralMsaParser.parse( is );
  88             }
  89             MsaCompactor mc = null;
  90             if ( worst_remove > 0 ) {
  91                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign );
  92             }
  93             else if ( av > 0 ) {
  94                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50 );
  95             }
  96             else if ( length > 0 ) {
  97                 mc = MsaCompactor.reduceLength( msa, length, step, realign );
  98             }
  99             System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
 100             for( final String id : mc.getRemovedSeqIds() ) {
 101                 System.out.println( id );
 102             }
 103             mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
 104         }
 105         catch ( final Exception e ) {
 106             e.printStackTrace();
 107             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 108         }
 109     }
 110
 111     private static void printHelp() {
 112         ForesterUtil.printProgramInformation( PRG_NAME,
 113                                               PRG_DESC,
 114                                               PRG_VERSION,
 115                                               PRG_DATE,
 116                                               E_MAIL,
 117                                               WWW,
 118                                               ForesterUtil.getForesterLibraryInformation() );
 119         System.out.println( "Usage:" );
 120         System.out.println();
 121         System.out.println( PRG_NAME + " <options> <msa input file>" );
 122         System.out.println();
 123         System.out.println( " options: " );
 124         System.out.println();
 125         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>: step size (msa columns)" );
 126         System.out.println();
 127         System.out.println();
 128         System.out.println();
 129     }
 130 }