forester/java/src/org/forester/archaeopteryx/tools/PhylogeneticInferrer.java

   1 // $Id:
   2 // forester -- software libraries and applications
   3 // for genomics and evolutionary biology research.
   4 //
   5 // Copyright (C) 2010 Christian M Zmasek
   6 // Copyright (C) 2010 Sanford-Burnham Medical Research Institute
   7 // All rights reserved
   8 //
   9 // This library is free software; you can redistribute it and/or
  10 // modify it under the terms of the GNU Lesser General Public
  11 // License as published by the Free Software Foundation; either
  12 // version 2.1 of the License, or (at your option) any later version.
  13 //
  14 // This library is distributed in the hope that it will be useful,
  15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 // Lesser General Public License for more details.
  18 //
  19 // You should have received a copy of the GNU Lesser General Public
  20 // License along with this library; if not, write to the Free Software
  21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  22 //
  23 // Contact: phylosoft @ gmail . com
  24 // WWW: www.phylosoft.org/forester
  25
  26 package org.forester.archaeopteryx.tools;
  27
  28 import java.io.BufferedWriter;
  29 import java.io.File;
  30 import java.io.FileWriter;
  31 import java.io.IOException;
  32 import java.util.ArrayList;
  33 import java.util.List;
  34 import java.util.regex.Matcher;
  35
  36 import javax.swing.JOptionPane;
  37
  38 import org.forester.archaeopteryx.AptxUtil;
  39 import org.forester.archaeopteryx.MainFrameApplication;
  40 import org.forester.evoinference.distance.NeighborJoining;
  41 import org.forester.evoinference.distance.PairwiseDistanceCalculator;
  42 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
  43 import org.forester.evoinference.tools.BootstrapResampler;
  44 import org.forester.io.parsers.FastaParser;
  45 import org.forester.msa.BasicMsa;
  46 import org.forester.msa.ClustalOmega;
  47 import org.forester.msa.Mafft;
  48 import org.forester.msa.Msa;
  49 import org.forester.msa.Msa.MSA_FORMAT;
  50 import org.forester.msa.MsaInferrer;
  51 import org.forester.msa.MsaMethods;
  52 import org.forester.msa.ResampleableMsa;
  53 import org.forester.phylogeny.Phylogeny;
  54 import org.forester.phylogeny.PhylogenyNode;
  55 import org.forester.phylogeny.data.Accession;
  56 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  57 import org.forester.sequence.Sequence;
  58 import org.forester.tools.ConfidenceAssessor;
  59 import org.forester.util.ForesterUtil;
  60
  61 public class PhylogeneticInferrer extends RunnableProcess {
  62
  63     private Msa                                _msa;
  64     private final MainFrameApplication         _mf;
  65     private final PhylogeneticInferenceOptions _options;
  66     private final List<Sequence>               _seqs;
  67     private final boolean                      DEBUG           = true;
  68     public final static String                 MSA_FILE_SUFFIX = ".aln";
  69     public final static String                 PWD_FILE_SUFFIX = ".pwd";
  70
  71     public PhylogeneticInferrer( final List<Sequence> seqs,
  72                                  final PhylogeneticInferenceOptions options,
  73                                  final MainFrameApplication mf ) {
  74         _msa = null;
  75         _seqs = seqs;
  76         _mf = mf;
  77         _options = options;
  78     }
  79
  80     public PhylogeneticInferrer( final Msa msa,
  81                                  final PhylogeneticInferenceOptions options,
  82                                  final MainFrameApplication mf ) {
  83         _msa = msa;
  84         _seqs = null;
  85         _mf = mf;
  86         _options = options;
  87     }
  88
  89     private Msa inferMsa() throws IOException, InterruptedException {
  90         //        final File temp_seqs_file = File.createTempFile( "__msa__temp__", ".fasta" );
  91         //        if ( DEBUG ) {
  92         //            System.out.println();
  93         //            System.out.println( "temp file: " + temp_seqs_file );
  94         //            System.out.println();
  95         //        }
  96         //        //final File temp_seqs_file = new File( _options.getTempDir() + ForesterUtil.FILE_SEPARATOR + "s.fasta" );
  97         //        final BufferedWriter writer = new BufferedWriter( new FileWriter( temp_seqs_file ) );
  98         //        SequenceWriter.writeSeqs( _seqs, writer, SEQ_FORMAT.FASTA, 100 );
  99         //        writer.close();
 100         final List<String> opts = processMafftOptions();
 101         return runMAFFT( _seqs, opts );
 102     }
 103
 104     private List<String> processMafftOptions() {
 105         final String opts_str = _options.getMsaPrgParameters().trim().toLowerCase();
 106         final String[] opts_ary = opts_str.split( " " );
 107         final List<String> opts = new ArrayList<String>();
 108         boolean saw_quiet = false;
 109         for( final String opt : opts_ary ) {
 110             opts.add( opt );
 111             if ( opt.equals( "--quiet" ) ) {
 112                 saw_quiet = true;
 113             }
 114         }
 115         if ( !saw_quiet ) {
 116             opts.add( "--quiet" );
 117         }
 118         return opts;
 119     }
 120
 121     private Phylogeny inferPhylogeny( final Msa msa ) {
 122         BasicSymmetricalDistanceMatrix m = null;
 123         switch ( _options.getPwdDistanceMethod() ) {
 124             case KIMURA_DISTANCE:
 125                 m = PairwiseDistanceCalculator.calcKimuraDistances( msa );
 126                 break;
 127             case POISSON_DISTANCE:
 128                 m = PairwiseDistanceCalculator.calcPoissonDistances( msa );
 129                 break;
 130             case FRACTIONAL_DISSIMILARITY:
 131                 m = PairwiseDistanceCalculator.calcFractionalDissimilarities( msa );
 132                 break;
 133             default:
 134                 throw new RuntimeException( "invalid pwd method" );
 135         }
 136         if ( !ForesterUtil.isEmpty( _options.getIntermediateFilesBase() ) ) {
 137             BufferedWriter pwd_writer;
 138             try {
 139                 pwd_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase() + PWD_FILE_SUFFIX ) );
 140                 m.write( pwd_writer );
 141                 pwd_writer.close();
 142             }
 143             catch ( final IOException e ) {
 144                 // TODO Auto-generated catch block
 145                 e.printStackTrace();
 146             }
 147         }
 148         final NeighborJoining nj = NeighborJoining.createInstance();
 149         final Phylogeny phy = nj.execute( m );
 150         PhylogeneticInferrer.extractFastaInformation( phy );
 151         return phy;
 152     }
 153
 154     private void infer() throws InterruptedException {
 155         //_mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
 156         if ( ( _msa == null ) && ( _seqs == null ) ) {
 157             throw new IllegalArgumentException( "cannot run phylogenetic analysis with null msa and seq array" );
 158         }
 159         start( _mf, "phylogenetic inference" );
 160         if ( _msa == null ) {
 161             Msa msa = null;
 162             try {
 163                 msa = inferMsa();
 164             }
 165             catch ( final IOException e ) {
 166                 end( _mf );
 167                 JOptionPane.showMessageDialog( _mf,
 168                                                "Could not create multiple sequence alignment with \""
 169                                                        + _options.getMsaPrg() + "\" and the following parameters:\n\""
 170                                                        + _options.getMsaPrgParameters() + "\"\nError: "
 171                                                        + e.getLocalizedMessage(),
 172                                                "Failed to Calculate MSA",
 173                                                JOptionPane.ERROR_MESSAGE );
 174                 if ( DEBUG ) {
 175                     e.printStackTrace();
 176                 }
 177                 return;
 178             }
 179             catch ( final Exception e ) {
 180                 end( _mf );
 181                 JOptionPane.showMessageDialog( _mf,
 182                                                "Could not create multiple sequence alignment with \""
 183                                                        + _options.getMsaPrg() + "\" and the following parameters:\n\""
 184                                                        + _options.getMsaPrgParameters() + "\"\nError: "
 185                                                        + e.getLocalizedMessage(),
 186                                                "Unexpected Exception During MSA Calculation",
 187                                                JOptionPane.ERROR_MESSAGE );
 188                 if ( DEBUG ) {
 189                     e.printStackTrace();
 190                 }
 191                 return;
 192             }
 193             if ( msa == null ) {
 194                 end( _mf );
 195                 JOptionPane.showMessageDialog( _mf,
 196                                                "Could not create multiple sequence alignment with "
 197                                                        + _options.getMsaPrg() + "\nand the following parameters:\n\""
 198                                                        + _options.getMsaPrgParameters() + "\"",
 199                                                "Failed to Calculate MSA",
 200                                                JOptionPane.ERROR_MESSAGE );
 201                 return;
 202             }
 203             if ( DEBUG ) {
 204                 System.out.println( msa.toString() );
 205                 System.out.println( MsaMethods.calcGapRatio( msa ) );
 206             }
 207             final MsaMethods msa_tools = MsaMethods.createInstance();
 208             if ( _options.isExecuteMsaProcessing() ) {
 209                 msa = msa_tools.removeGapColumns( _options.getMsaProcessingMaxAllowedGapRatio(),
 210                                                   _options.getMsaProcessingMinAllowedLength(),
 211                                                   msa );
 212                 if ( msa == null ) {
 213                     end( _mf );
 214                     JOptionPane.showMessageDialog( _mf,
 215                                                    "Less than two sequences longer than "
 216                                                            + _options.getMsaProcessingMinAllowedLength()
 217                                                            + " residues left after MSA processing",
 218                                                    "MSA Processing Settings Too Stringent",
 219                                                    JOptionPane.ERROR_MESSAGE );
 220                     return;
 221                 }
 222             }
 223             if ( DEBUG ) {
 224                 System.out.println( msa_tools.getIgnoredSequenceIds() );
 225                 System.out.println( msa.toString() );
 226                 System.out.println( MsaMethods.calcGapRatio( msa ) );
 227             }
 228             _msa = msa;
 229         }
 230         final int n = _options.getBootstrapSamples();
 231         final long seed = _options.getRandomNumberGeneratorSeed();
 232         final Phylogeny master_phy = inferPhylogeny( _msa );
 233         if ( _options.isPerformBootstrapResampling() && ( n > 0 ) ) {
 234             final ResampleableMsa resampleable_msa = new ResampleableMsa( ( BasicMsa ) _msa );
 235             final int[][] resampled_column_positions = BootstrapResampler.createResampledColumnPositions( _msa
 236                     .getLength(), n, seed );
 237             final Phylogeny[] eval_phys = new Phylogeny[ n ];
 238             for( int i = 0; i < n; ++i ) {
 239                 resampleable_msa.resample( resampled_column_positions[ i ] );
 240                 eval_phys[ i ] = inferPhylogeny( resampleable_msa );
 241             }
 242             ConfidenceAssessor.evaluate( "bootstrap", eval_phys, master_phy, true, 1 );
 243         }
 244         _mf.getMainPanel().addPhylogenyInNewTab( master_phy, _mf.getConfiguration(), "nj", "njpath" );
 245         //  _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
 246         end( _mf );
 247         JOptionPane.showMessageDialog( _mf,
 248                                        "Inference successfully completed",
 249                                        "Inference Completed",
 250                                        JOptionPane.INFORMATION_MESSAGE );
 251     }
 252
 253     @Override
 254     public void run() {
 255         try {
 256             infer();
 257         }
 258         catch ( final InterruptedException e ) {
 259             // TODO need to handle this exception SOMEHOW!
 260             // TODO Auto-generated catch block
 261             e.printStackTrace();
 262         }
 263     }
 264
 265     private Msa runMAFFT( final List<Sequence> seqs, final List<String> opts ) throws IOException, InterruptedException {
 266         Msa msa = null;
 267         final MsaInferrer mafft = Mafft.createInstance( _mf.getInferenceManager().getPathToLocalMafft()
 268                 .getCanonicalPath() );
 269         try {
 270             msa = mafft.infer( seqs, opts );
 271         }
 272         catch ( final IOException e ) {
 273             System.out.println( mafft.getErrorDescription() );
 274         }
 275         return msa;
 276     }
 277
 278     private Msa runClustalOmega( final File input_seqs, final List<String> opts ) throws IOException,
 279             InterruptedException {
 280         Msa msa = null;
 281         final MsaInferrer clustalo = ClustalOmega.createInstance( _mf.getInferenceManager().getPathToLocalClustalo()
 282                 .getCanonicalPath() );
 283         try {
 284             msa = clustalo.infer( input_seqs, opts );
 285         }
 286         catch ( final IOException e ) {
 287             System.out.println( clustalo.getErrorDescription() );
 288         }
 289         return msa;
 290     }
 291
 292     private void writeToFiles( final BasicSymmetricalDistanceMatrix m ) {
 293         if ( !ForesterUtil.isEmpty( _options.getIntermediateFilesBase() ) ) {
 294             try {
 295                 final BufferedWriter msa_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase()
 296                         + MSA_FILE_SUFFIX ) );
 297                 _msa.write( msa_writer, MSA_FORMAT.PHYLIP );
 298                 msa_writer.close();
 299                 final BufferedWriter pwd_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase()
 300                         + PWD_FILE_SUFFIX ) );
 301                 m.write( pwd_writer );
 302                 pwd_writer.close();
 303             }
 304             catch ( final Exception e ) {
 305                 System.out.println( "Error: " + e.getMessage() );
 306             }
 307         }
 308     }
 309
 310     public static void extractFastaInformation( final Phylogeny phy ) {
 311         for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
 312             final PhylogenyNode node = iter.next();
 313             if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 314                 final Matcher name_m = FastaParser.FASTA_DESC_LINE.matcher( node.getName() );
 315                 if ( name_m.lookingAt() ) {
 316                     System.out.println();
 317                     // System.out.println( name_m.group( 1 ) );
 318                     // System.out.println( name_m.group( 2 ) );
 319                     // System.out.println( name_m.group( 3 ) );
 320                     // System.out.println( name_m.group( 4 ) );
 321                     final String acc_source = name_m.group( 1 );
 322                     final String acc = name_m.group( 2 );
 323                     final String seq_name = name_m.group( 3 );
 324                     final String tax_sn = name_m.group( 4 );
 325                     if ( !ForesterUtil.isEmpty( acc_source ) && !ForesterUtil.isEmpty( acc ) ) {
 326                         AptxUtil.ensurePresenceOfSequence( node );
 327                         node.getNodeData().getSequence( 0 ).setAccession( new Accession( acc, acc_source ) );
 328                     }
 329                     if ( !ForesterUtil.isEmpty( seq_name ) ) {
 330                         AptxUtil.ensurePresenceOfSequence( node );
 331                         node.getNodeData().getSequence( 0 ).setName( seq_name );
 332                     }
 333                     if ( !ForesterUtil.isEmpty( tax_sn ) ) {
 334                         AptxUtil.ensurePresenceOfTaxonomy( node );
 335                         node.getNodeData().getTaxonomy( 0 ).setScientificName( tax_sn );
 336                     }
 337                 }
 338             }
 339         }
 340     }
 341 }