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.MsaCompactor;
  13 import org.forester.msa.MsaMethods;
  14 import org.forester.util.CommandLineArguments;
  15 import org.forester.util.ForesterUtil;
  16
  17 public class msa_compactor {
  18
  19     final static private String HELP_OPTION_1                 = "help";
  20     final static private String HELP_OPTION_2                 = "h";
  21     final static private String REMOVE_WORST_OFFENDERS_OPTION = "w";
  22     final static private String AV_GAPINESS_OPTION            = "a";
  23     final static private String LENGTH_OPTION                 = "l";
  24     final static private String REALIGN_OPTION                = "r";
  25     final static private String PRG_NAME                      = "msa_compactor";
  26     final static private String PRG_DESC                      = "multiple sequnce aligment compactor";
  27     final static private String PRG_VERSION                   = "0.90";
  28     final static private String PRG_DATE                      = "2012.07.11";
  29     final static private String E_MAIL                        = "phylosoft@gmail.com";
  30     final static private String WWW                           = "www.phylosoft.org/forester/";
  31
  32     public static void main( final String args[] ) {
  33         try {
  34             final CommandLineArguments cla = new CommandLineArguments( args );
  35             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) ) {
  36                 printHelp();
  37                 System.exit( 0 );
  38             }
  39             final File in = cla.getFile( 0 );
  40             int worst_remove = -1;
  41             double av = -1;
  42             int length = -1;
  43             final int step = 5;
  44             boolean realign = false;
  45             //            int to = 0;
  46             //            int window = 0;
  47             //            int step = 0;
  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             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  54             if ( dissallowed_options.length() > 0 ) {
  55                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  56             }
  57             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  58                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  59             }
  60             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  61                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  62             }
  63             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  64                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  65             }
  66             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  67                 realign = true;
  68             }
  69             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  70             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  71             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  72             //            }
  73             //            else {
  74             //                printHelp();
  75             //                System.exit( 0 );
  76             //            }
  77             Msa msa = null;
  78             final FileInputStream is = new FileInputStream( in );
  79             if ( FastaParser.isLikelyFasta( in ) ) {
  80                 msa = FastaParser.parseMsa( is );
  81             }
  82             else {
  83                 msa = GeneralMsaParser.parse( is );
  84             }
  85             System.out.println( msa.toString() );
  86             System.out.println( MsaMethods.calcBasicGapinessStatistics( msa ).arithmeticMean() );
  87             MsaCompactor mc = null;
  88             if ( worst_remove > 0 ) {
  89                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign );
  90             }
  91             else if ( av > 0 ) {
  92                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign );
  93             }
  94             else if ( length > 0 ) {
  95                 mc = MsaCompactor.reduceLength( msa, length, step, realign );
  96             }
  97             System.out.println( mc.getMsa().toString() );
  98             System.out.println( MsaMethods.calcBasicGapinessStatistics( mc.getMsa() ).arithmeticMean() );
  99             for( final String id : mc.getRemovedSeqIds() ) {
 100                 System.out.println( id );
 101             }
 102         }
 103         catch ( final Exception e ) {
 104             e.printStackTrace();
 105             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 106         }
 107     }
 108
 109     private static void printHelp() {
 110         ForesterUtil.printProgramInformation( PRG_NAME,
 111                                               PRG_DESC,
 112                                               PRG_VERSION,
 113                                               PRG_DATE,
 114                                               E_MAIL,
 115                                               WWW,
 116                                               ForesterUtil.getForesterLibraryInformation() );
 117         System.out.println( "Usage:" );
 118         System.out.println();
 119         System.out.println( PRG_NAME + " <options> <msa input file>" );
 120         System.out.println();
 121         System.out.println( " options: " );
 122         System.out.println();
 123         //        System.out.println( "   -" + FROM_OPTION + "=<integer>: from (msa column)" );
 124         //        System.out.println( "   -" + TO_OPTION + "=<integer>: to (msa column)" );
 125         //        System.out.println( "    or" );
 126         //        System.out.println( "   -" + WINDOW_OPTION + "=<integer>: window size (msa columns)" );
 127         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>: step size (msa columns)" );
 128         System.out.println();
 129         System.out.println();
 130         System.out.println();
 131     }
 132 }