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.MsaInferrer;
  13 import org.forester.msa.Msa.MSA_FORMAT;
  14 import org.forester.msa.MsaMethods;
  15 import org.forester.msa_compactor.MsaCompactor;
  16 import org.forester.util.CommandLineArguments;
  17 import org.forester.util.ForesterUtil;
  18
  19 public class msa_compactor {
  20
  21     final static private String HELP_OPTION_1                          = "help";
  22     final static private String HELP_OPTION_2                          = "h";
  23     final static private String REMOVE_WORST_OFFENDERS_OPTION          = "w";
  24     final static private String AV_GAPINESS_OPTION                     = "a";
  25     final static private String STEP_OPTION                            = "s";
  26     final static private String LENGTH_OPTION                          = "l";
  27     final static private String REALIGN_OPTION                         = "r";
  28     final static private String PATH_TO_MAFFT_OPTION                   = "mafft";
  29     final static private String DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION = "nn";
  30     final static private String PRG_NAME                               = "msa_compactor";
  31     final static private String PRG_DESC                               = "multiple sequnce aligment compactor";
  32     final static private String PRG_VERSION                            = "0.01";
  33     final static private String PRG_DATE                               = "140314";
  34     final static private String E_MAIL                                 = "phylosoft@gmail.com";
  35     final static private String WWW                                    = "https://sites.google.com/site/cmzmasek/home/software/forester";
  36
  37     public static void main( final String args[] ) {
  38         try {
  39             final CommandLineArguments cla = new CommandLineArguments( args );
  40             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
  41                 printHelp();
  42                 System.exit( 0 );
  43             }
  44             final File in = cla.getFile( 0 );
  45             final File out = cla.getFile( 1 );
  46             int worst_remove = -1;
  47             double av = -1;
  48             int length = -1;
  49             int step = 1;
  50             boolean realign = false;
  51             boolean norm = true;
  52             String path_to_mafft = null;
  53             final List<String> allowed_options = new ArrayList<String>();
  54             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
  55             allowed_options.add( AV_GAPINESS_OPTION );
  56             allowed_options.add( LENGTH_OPTION );
  57             allowed_options.add( REALIGN_OPTION );
  58             allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
  59             allowed_options.add( STEP_OPTION );
  60             allowed_options.add( PATH_TO_MAFFT_OPTION );
  61             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  62             if ( dissallowed_options.length() > 0 ) {
  63                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  64             }
  65             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  66                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  67             }
  68             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  69                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  70             }
  71             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  72                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  73             }
  74             if ( cla.isOptionSet( STEP_OPTION ) ) {
  75                 step = cla.getOptionValueAsInt( STEP_OPTION );
  76             }
  77             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  78                 realign = true;
  79             }
  80             if ( cla.isOptionSet( PATH_TO_MAFFT_OPTION ) ) {
  81                 if ( !realign ) {
  82                     ForesterUtil.fatalError( PRG_NAME, "no need to indicate path to MAFFT without realigning" );
  83                 }
  84                 path_to_mafft = cla.getOptionValueAsCleanString( PATH_TO_MAFFT_OPTION );
  85             }
  86             if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
  87                 norm = false;
  88             }
  89             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  90             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  91             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  92             //            }
  93             //            else {
  94             //                printHelp();
  95             //                System.exit( 0 );
  96             //            }
  97             Msa msa = null;
  98             final FileInputStream is = new FileInputStream( in );
  99             if ( FastaParser.isLikelyFasta( in ) ) {
 100                 msa = FastaParser.parseMsa( is );
 101             }
 102             else {
 103                 msa = GeneralMsaParser.parse( is );
 104             }
 105
 106             if ( realign ) {
 107                 if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
 108                     path_to_mafft = MsaCompactor.guessPathToMafft();
 109                 }
 110                 if ( !ForesterUtil.isEmpty( path_to_mafft ) && MsaInferrer.isInstalled( path_to_mafft ) ) {
 111                     ForesterUtil.programMessage( PRG_NAME, "using MAFFT at \"" + path_to_mafft + "\""  );
 112                 }
 113                 else {
 114                     if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
 115                         ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable found, use -\"" + PATH_TO_MAFFT_OPTION + "=<path to MAFFT>\" option" );
 116                     }
 117                     else {
 118                         ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable at \""  + path_to_mafft + "\""  );
 119                     }
 120                 }
 121             }
 122
 123             MsaCompactor mc = null;
 124             if ( worst_remove > 0 ) {
 125                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign, norm );
 126             }
 127             else if ( av > 0 ) {
 128                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50 );
 129             }
 130             else if ( length > 0 ) {
 131                 mc = MsaCompactor.reduceLength( msa, length, step, realign );
 132             }
 133             System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
 134             for( final String id : mc.getRemovedSeqIds() ) {
 135                 System.out.println( id );
 136             }
 137             mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
 138         }
 139         catch ( final Exception e ) {
 140             e.printStackTrace();
 141             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 142         }
 143     }
 144
 145     private static void printHelp() {
 146         ForesterUtil.printProgramInformation( PRG_NAME,
 147                                               PRG_DESC,
 148                                               PRG_VERSION,
 149                                               PRG_DATE,
 150                                               E_MAIL,
 151                                               WWW,
 152                                               ForesterUtil.getForesterLibraryInformation() );
 153         final String path_to_mafft = MsaCompactor.guessPathToMafft();
 154         String mafft_comment;
 155         if ( !ForesterUtil.isEmpty( path_to_mafft ) ) {
 156             mafft_comment = " (" + path_to_mafft + ")";
 157         }
 158         else {
 159             mafft_comment = " (no path to MAFFT found, use -\"" + PATH_TO_MAFFT_OPTION + "=<path to MAFFT>\" option";
 160         }
 161
 162         System.out.println( "Usage:" );
 163         System.out.println();
 164         System.out.println( PRG_NAME + " <options> <msa input file>" );
 165         System.out.println();
 166         System.out.println( " options: " );
 167         System.out.println();
 168         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION + "=<integer>  number of sequences to remove" );
 169         System.out.println( "   -" + REALIGN_OPTION + " to realign using MAFFT" + mafft_comment );
 170         System.out.println();
 171         System.out.println();
 172         System.out.println();
 173     }
 174 }