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