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.MsaInferrer;
  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          = "r";
  23     final static private String AV_GAPINESS_OPTION                     = "g";
  24     final static private String STEP_OPTION                            = "s";
  25     final static private String LENGTH_OPTION                          = "l";
  26     final static private String REALIGN_OPTION                         = "a";
  27     final static private String PATH_TO_MAFFT_OPTION                   = "mafft";
  28     final static private String DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION = "nn";
  29     final static private String PRG_NAME                               = "msa_compactor";
  30     final static private String PRG_DESC                               = "multiple sequnce aligment compactor";
  31     final static private String PRG_VERSION                            = "0.01";
  32     final static private String PRG_DATE                               = "140314";
  33     final static private String E_MAIL                                 = "phylosoft@gmail.com";
  34     final static private String WWW                                    = "https://sites.google.com/site/cmzmasek/home/software/forester";
  35
  36     public static void main( final String args[] ) {
  37         try {
  38             final CommandLineArguments cla = new CommandLineArguments( args );
  39             if ( cla.isOptionSet( HELP_OPTION_1 ) || cla.isOptionSet( HELP_OPTION_2 ) || ( cla.getNumberOfNames() != 2 ) ) {
  40                 printHelp();
  41                 System.exit( 0 );
  42             }
  43             final File in = cla.getFile( 0 );
  44             final File out = cla.getFile( 1 );
  45             int worst_remove = -1;
  46             double av = -1;
  47             int length = -1;
  48             int step = 1;
  49             boolean realign = false;
  50             boolean norm = true;
  51             String path_to_mafft = null;
  52             final List<String> allowed_options = new ArrayList<String>();
  53             allowed_options.add( REMOVE_WORST_OFFENDERS_OPTION );
  54             allowed_options.add( AV_GAPINESS_OPTION );
  55             allowed_options.add( LENGTH_OPTION );
  56             allowed_options.add( REALIGN_OPTION );
  57             allowed_options.add( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION );
  58             allowed_options.add( STEP_OPTION );
  59             allowed_options.add( PATH_TO_MAFFT_OPTION );
  60             final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
  61             if ( dissallowed_options.length() > 0 ) {
  62                 ForesterUtil.fatalError( PRG_NAME, "unknown option(s): " + dissallowed_options );
  63             }
  64             if ( cla.isOptionSet( REMOVE_WORST_OFFENDERS_OPTION ) ) {
  65                 worst_remove = cla.getOptionValueAsInt( REMOVE_WORST_OFFENDERS_OPTION );
  66             }
  67             if ( cla.isOptionSet( AV_GAPINESS_OPTION ) ) {
  68                 av = cla.getOptionValueAsDouble( AV_GAPINESS_OPTION );
  69             }
  70             if ( cla.isOptionSet( LENGTH_OPTION ) ) {
  71                 length = cla.getOptionValueAsInt( LENGTH_OPTION );
  72             }
  73             if ( cla.isOptionSet( STEP_OPTION ) ) {
  74                 step = cla.getOptionValueAsInt( STEP_OPTION );
  75             }
  76             if ( cla.isOptionSet( REALIGN_OPTION ) ) {
  77                 realign = true;
  78             }
  79             if ( cla.isOptionSet( PATH_TO_MAFFT_OPTION ) ) {
  80                 if ( !realign ) {
  81                     ForesterUtil.fatalError( PRG_NAME, "no need to indicate path to MAFFT without realigning" );
  82                 }
  83                 path_to_mafft = cla.getOptionValueAsCleanString( PATH_TO_MAFFT_OPTION );
  84             }
  85             if ( cla.isOptionSet( DO_NOT_NORMALIZE_FOR_EFF_LENGTH_OPTION ) ) {
  86                 norm = false;
  87             }
  88             //            else if ( cla.isOptionSet( STEP_OPTION ) && cla.isOptionSet( WINDOW_OPTION ) ) {
  89             //                step = cla.getOptionValueAsInt( STEP_OPTION );
  90             //                window = cla.getOptionValueAsInt( WINDOW_OPTION );
  91             //            }
  92             //            else {
  93             //                printHelp();
  94             //                System.exit( 0 );
  95             //            }
  96             if ( realign ) {
  97                 if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
  98                     path_to_mafft = MsaCompactor.guessPathToMafft();
  99                 }
 100                 checkPathToMafft( path_to_mafft );
 101             }
 102             Msa msa = null;
 103             final FileInputStream is = new FileInputStream( in );
 104             if ( FastaParser.isLikelyFasta( in ) ) {
 105                 msa = FastaParser.parseMsa( is );
 106             }
 107             else {
 108                 msa = GeneralMsaParser.parse( is );
 109             }
 110             MsaCompactor mc = null;
 111             if ( worst_remove > 0 ) {
 112                 mc = MsaCompactor.removeWorstOffenders( msa, worst_remove, realign, norm, path_to_mafft );
 113             }
 114             else if ( av > 0 ) {
 115                 mc = MsaCompactor.reduceGapAverage( msa, av, step, realign, out, 50, path_to_mafft );
 116             }
 117             else if ( length > 0 ) {
 118                 mc = MsaCompactor.reduceLength( msa, length, step, realign, path_to_mafft );
 119             }
 120             //System.out.println( MsaMethods.calcGapRatio( mc.getMsa() ) );
 121             // for( final String id : mc.getRemovedSeqIds() ) {
 122             //     System.out.println( id );
 123             //}
 124             mc.writeMsa( out, MSA_FORMAT.PHYLIP, ".aln" );
 125         }
 126         catch ( final Exception e ) {
 127             e.printStackTrace();
 128             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
 129         }
 130     }
 131
 132     private static void checkPathToMafft( final String path_to_mafft ) {
 133         if ( !ForesterUtil.isEmpty( path_to_mafft ) && MsaInferrer.isInstalled( path_to_mafft ) ) {
 134             ForesterUtil.programMessage( PRG_NAME, "using MAFFT at \"" + path_to_mafft + "\"" );
 135         }
 136         else {
 137             if ( ForesterUtil.isEmpty( path_to_mafft ) ) {
 138                 ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable found, use -\"" + PATH_TO_MAFFT_OPTION
 139                         + "=<path to MAFFT>\" option" );
 140             }
 141             else {
 142                 ForesterUtil.fatalError( PRG_NAME, "no MAFFT executable at \"" + path_to_mafft + "\"" );
 143             }
 144         }
 145     }
 146
 147     private static void printHelp() {
 148         ForesterUtil.printProgramInformation( PRG_NAME,
 149                                               PRG_DESC,
 150                                               PRG_VERSION,
 151                                               PRG_DATE,
 152                                               E_MAIL,
 153                                               WWW,
 154                                               ForesterUtil.getForesterLibraryInformation() );
 155         final String path_to_mafft = MsaCompactor.guessPathToMafft();
 156         String mafft_comment;
 157         if ( !ForesterUtil.isEmpty( path_to_mafft ) ) {
 158             mafft_comment = " (using " + path_to_mafft + ")";
 159         }
 160         else {
 161             mafft_comment = " (no path to MAFFT found, use -\"" + PATH_TO_MAFFT_OPTION + "=<path to MAFFT>\" option";
 162         }
 163         System.out.println( "Usage:" );
 164         System.out.println();
 165         System.out.println( PRG_NAME + " <options> <msa input file> <output file>" );
 166         System.out.println();
 167         System.out.println( " options: " );
 168         System.out.println();
 169         System.out.println( "   -" + REMOVE_WORST_OFFENDERS_OPTION
 170                 + "=<integer>  number of worst offender sequences to remove" );
 171         System.out.println( "   -" + REALIGN_OPTION + "            to realign using MAFFT" + mafft_comment );
 172         System.out.println();
 173         System.out.println();
 174         System.out.println();
 175     }
 176 }