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_compactor.MsaCompactor;
  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          = "r";
  22     final static private String AV_GAPINESS_OPTION                     = "g";
  23     final static private String STEP_OPTION                            = "s";
  24     final static private String LENGTH_OPTION                          = "l";
  25     final static private String REALIGN_OPTION                         = "a";
  26     final static private String PATH_TO_MAFFT_OPTION                   = "mafft";
  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             String path_to_mafft = null;
  51             final List<String> allowed_options = new ArrayList<String>();
  52             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
  53             allowed_options.add( AV_GAPINESS_OPTION );
  54             allowed_options.add( LENGTH_OPTION );
  55             allowed_options.add( REALIGN_OPTION );
  56             allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
  57             allowed_options.add( STEP_OPTION );
  58             allowed_options.add( PATH_TO_MAFFT_OPTION );
  59             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  60             if ( dissallowed_options.length() > 0 ) {
  61                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  62             }
  63             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  64                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  65             }
  66             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  67                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  68             }
  69             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  70                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  71             }
  72             if ( cla.isOptionSet( STEP_OPTION ) ) {
  73                 step = cla.getOptionValueAsInt( STEP_OPTION );
  74             }
  75             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  76                 realign = true;
  77             }
  78             if ( cla.isOptionSet( PATH_TO_MAFFT_OPTION ) ) {
  79                 if ( !realign ) {
  80                     ForesterUtil.fatalError( PRG_NAME, "no need to indicate path to MAFFT without realigning" );
  81                 }
  82                 path_to_mafft = cla.getOptionValueAsCleanString( PATH_TO_MAFFT_OPTION );
  83             }
  84             if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
  85                 norm = false;
  86             }
  87             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  88             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  89             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  90             //            }
  91             //            else {
  92             //                printHelp();
  93             //                System.exit( 0 );
  94             //            }
  95             if ( realign ) {
  96                 if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
  97                     path_to_mafft = MsaCompactor.guessPathToMafft();
  98                 }
  99                 checkPathToMafft( path_to_mafft );
 100             }
 101             Msa msa = null;
 102             final FileInputStream is = new FileInputStream( in );
 103             if ( FastaParser.isLikelyFasta( in ) ) {
 104                 msa = FastaParser.parseMsa( is );
 105             }
 106             else {
 107                 msa = GeneralMsaParser.parse( is );
 108             }
 109             MsaCompactor mc = null;
 110             if ( worst_remove > 0 ) {
 111                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, step, realign, norm, path_to_mafft, out );
 112             }
 113             else if ( av > 0 ) {
 114                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, norm, path_to_mafft, out );
 115             }
 116             else if ( length > 0 ) {
 117                 if ( length >= msa.getLength() ) {
 118                     ForesterUtil.fatalError( PRG_NAME, "target MSA length (" + length
 119                             + ") is greater than or equal to MSA original length (" + msa.getLength() + ")" );
 120                 }
 121                 // TODO if < shortest seq -> error
 122                 mc = MsaCompactor.reduceLength( msa, length, step, realign, norm, path_to_mafft, out );
 123             }
 124             //System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
 125             // for( final String id : mc.getRemovedSeqIds() ) {
 126             //     System.out.println( id );
 127             //}
 128             //mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
 129         }
 130         catch ( final Exception e ) {
 131             e.printStackTrace();
 132             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 133         }
 134     }
 135
 136     private static void checkPathToMafft( final String path_to_mafft ) {
 137         if ( !ForesterUtil.isEmpty( path_to_mafft ) && MsaInferrer.isInstalled( path_to_mafft ) ) {
 138             ForesterUtil.programMessage( PRG_NAME, "using MAFFT at \"" + path_to_mafft + "\"" );
 139         }
 140         else {
 141             if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
 142                 ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable found, use -\"" + PATH_TO_MAFFT_OPTION
 143                         + "=<path to MAFFT>\" option" );
 144             }
 145             else {
 146                 ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable at \"" + path_to_mafft + "\"" );
 147             }
 148         }
 149     }
 150
 151     private static void printHelp() {
 152         ForesterUtil.printProgramInformation( PRG_NAME,
 153                                               PRG_DESC,
 154                                               PRG_VERSION,
 155                                               PRG_DATE,
 156                                               E_MAIL,
 157                                               WWW,
 158                                               ForesterUtil.getForesterLibraryInformation() );
 159         final String path_to_mafft = MsaCompactor.guessPathToMafft();
 160         String mafft_comment;
 161         if ( !ForesterUtil.isEmpty( path_to_mafft ) ) {
 162             mafft_comment = " (using " + path_to_mafft + ")";
 163         }
 164         else {
 165             mafft_comment = " (no path to MAFFT found, use -\"" + PATH_TO_MAFFT_OPTION + "=<path to MAFFT>\" option";
 166         }
 167         System.out.println( "Usage:" );
 168         System.out.println();
 169         System.out.println( PRG_NAME + " <options> <msa input file> <output file>" );
 170         System.out.println();
 171         System.out.println( " options: " );
 172         System.out.println();
 173         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION
 174                 + "=<integer>  number of worst offender sequences to remove" );
 175         System.out.println( "   -" + LENGTH_OPTION + "=<integer>  target MSA length" );
 176         System.out.println( "   -" + AV_GAPINESS_OPTION + "=<decimal>  gap %" );
 177         System.out.println( "   -" + STEP_OPTION + "=<decimal>  step" );
 178         System.out.println( "   -" + REALIGN_OPTION + "            to realign using MAFFT" + mafft_comment );
 179         System.out.println();
 180         System.out.println();
 181         System.out.println();
 182     }
 183 }