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.io.writers.SequenceWriter;
  46 import org.forester.io.writers.SequenceWriter.SEQ_FORMAT;
  47 import org.forester.msa.BasicMsa;
  48 import org.forester.msa.Mafft;
  49 import org.forester.msa.Msa;
  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 implements Runnable {
  62
  63     private Msa                                _msa;
  64     private final MainFrameApplication         _mf;
  65     private final PhylogeneticInferenceOptions _options;
  66     private final List<Sequence>               _seqs;
  67     public final static String                 MSA_FILE_SUFFIX = ".aln";
  68     public final static String                 PWD_FILE_SUFFIX = ".pwd";
  69
  70     public PhylogeneticInferrer( final List<Sequence> seqs,
  71                                  final PhylogeneticInferenceOptions options,
  72                                  final MainFrameApplication mf ) {
  73         _msa = null;
  74         _seqs = seqs;
  75         _mf = mf;
  76         _options = options;
  77     }
  78
  79     public PhylogeneticInferrer( final Msa msa,
  80                                  final PhylogeneticInferenceOptions options,
  81                                  final MainFrameApplication mf ) {
  82         _msa = msa;
  83         _seqs = null;
  84         _mf = mf;
  85         _options = options;
  86     }
  87
  88     private Msa inferMsa() throws IOException, InterruptedException {
  89         final File temp_seqs_file = File.createTempFile( "__msa__temp__", ".fasta" );
  90         System.out.println();
  91         System.out.println( "temp file: " + temp_seqs_file );
  92         System.out.println();
  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         final List<String> opts = processMafftOptions();
  98         return runMAFFT( temp_seqs_file, opts );
  99     }
 100
 101     private List<String> processMafftOptions() {
 102         final String opts_str = _options.getMsaPrgParameters().trim().toLowerCase();
 103         final String[] opts_ary = opts_str.split( " " );
 104         final List<String> opts = new ArrayList<String>();
 105         boolean saw_quiet = false;
 106         for( final String opt : opts_ary ) {
 107             opts.add( opt );
 108             if ( opt.equals( "--quiet" ) ) {
 109                 saw_quiet = true;
 110             }
 111         }
 112         if ( !saw_quiet ) {
 113             opts.add( "--quiet" );
 114         }
 115         return opts;
 116     }
 117
 118     private Phylogeny inferPhylogeny( final Msa msa ) {
 119         BasicSymmetricalDistanceMatrix m = null;
 120         switch ( _options.getPwdDistanceMethod() ) {
 121             case KIMURA_DISTANCE:
 122                 m = PairwiseDistanceCalculator.calcKimuraDistances( msa );
 123                 break;
 124             case POISSON_DISTANCE:
 125                 m = PairwiseDistanceCalculator.calcPoissonDistances( msa );
 126                 break;
 127             case FRACTIONAL_DISSIMILARITY:
 128                 m = PairwiseDistanceCalculator.calcFractionalDissimilarities( msa );
 129                 break;
 130             default:
 131                 throw new RuntimeException( "invalid pwd method" );
 132         }
 133         if ( !ForesterUtil.isEmpty( _options.getIntermediateFilesBase() ) ) {
 134             BufferedWriter pwd_writer;
 135             try {
 136                 pwd_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase() + PWD_FILE_SUFFIX ) );
 137                 m.write( pwd_writer );
 138                 pwd_writer.close();
 139             }
 140             catch ( final IOException e ) {
 141                 // TODO Auto-generated catch block
 142                 e.printStackTrace();
 143             }
 144         }
 145         final NeighborJoining nj = new NeighborJoining();
 146         final Phylogeny phy = nj.execute( m );
 147         PhylogeneticInferrer.extractFastaInformation( phy );
 148         return phy;
 149     }
 150
 151     private void infer() throws InterruptedException {
 152         //_mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
 153         if ( ( _msa == null ) && ( _seqs == null ) ) {
 154             throw new IllegalArgumentException( "cannot run phylogenetic analysis with null msa and seq array" );
 155         }
 156         if ( _msa == null ) {
 157             Msa msa = null;
 158             try {
 159                 msa = inferMsa();
 160             }
 161             catch ( final IOException e ) {
 162                 JOptionPane.showMessageDialog( _mf,
 163                                                "Could not create multiple sequence alignment with "
 164                                                        + _options.getMsaPrg() + "\nand the following parameters:\n\""
 165                                                        + _options.getMsaPrgParameters() + "\"\nError:"
 166                                                        + e.getLocalizedMessage(),
 167                                                "Failed to Calculate MSA",
 168                                                JOptionPane.ERROR_MESSAGE );
 169                 return;
 170             }
 171             if ( msa == null ) {
 172                 JOptionPane.showMessageDialog( _mf,
 173                                                "Could not create multiple sequence alignment with "
 174                                                        + _options.getMsaPrg() + "\nand the following parameters:\n\""
 175                                                        + _options.getMsaPrgParameters() + "\"",
 176                                                "Failed to Calculate MSA",
 177                                                JOptionPane.ERROR_MESSAGE );
 178                 return;
 179             }
 180             System.out.println( msa.toString() );
 181             System.out.println( MsaMethods.calcBasicGapinessStatistics( msa ).toString() );
 182             final MsaMethods msa_tools = MsaMethods.createInstance();
 183             if ( _options.isExecuteMsaProcessing() ) {
 184                 msa = msa_tools.removeGapColumns( _options.getMsaProcessingMaxAllowedGapRatio(),
 185                                                   _options.getMsaProcessingMinAllowedLength(),
 186                                                   msa );
 187                 if ( msa == null ) {
 188                     JOptionPane.showMessageDialog( _mf,
 189                                                    "Less than two sequences longer than "
 190                                                            + _options.getMsaProcessingMinAllowedLength()
 191                                                            + " residues left after MSA processing",
 192                                                    "MSA Processing Settings Too Stringent",
 193                                                    JOptionPane.ERROR_MESSAGE );
 194                     return;
 195                 }
 196             }
 197             System.out.println( msa_tools.getIgnoredSequenceIds() );
 198             System.out.println( msa.toString() );
 199             System.out.println( MsaMethods.calcBasicGapinessStatistics( msa ).toString() );
 200             _msa = msa;
 201         }
 202         final int n = _options.getBootstrapSamples();
 203         final long seed = _options.getRandomNumberGeneratorSeed();
 204         final Phylogeny master_phy = inferPhylogeny( _msa );
 205         if ( _options.isPerformBootstrapResampling() && ( n > 0 ) ) {
 206             final ResampleableMsa resampleable_msa = new ResampleableMsa( ( BasicMsa ) _msa );
 207             final int[][] resampled_column_positions = BootstrapResampler.createResampledColumnPositions( _msa
 208                     .getLength(), n, seed );
 209             final Phylogeny[] eval_phys = new Phylogeny[ n ];
 210             for( int i = 0; i < n; ++i ) {
 211                 resampleable_msa.resample( resampled_column_positions[ i ] );
 212                 eval_phys[ i ] = inferPhylogeny( resampleable_msa );
 213             }
 214             ConfidenceAssessor.evaluate( "bootstrap", eval_phys, master_phy, true, 1 );
 215         }
 216         _mf.getMainPanel().addPhylogenyInNewTab( master_phy, _mf.getConfiguration(), "nj", "njpath" );
 217         _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
 218         JOptionPane.showMessageDialog( _mf,
 219                                        "Inference successfully completed",
 220                                        "Inference Completed",
 221                                        JOptionPane.INFORMATION_MESSAGE );
 222     }
 223
 224     @Override
 225     public void run() {
 226         try {
 227             infer();
 228         }
 229         catch ( final InterruptedException e ) {
 230             // TODO need to handle this exception SOMEHOW!
 231             // TODO Auto-generated catch block
 232             e.printStackTrace();
 233         }
 234     }
 235
 236     private Msa runMAFFT( final File input_seqs, final List<String> opts ) throws IOException, InterruptedException {
 237         Msa msa = null;
 238         final MsaInferrer mafft = Mafft.createInstance();
 239         try {
 240             msa = mafft.infer( input_seqs, opts );
 241         }
 242         catch ( final IOException e ) {
 243             System.out.println( mafft.getErrorDescription() );
 244         }
 245         return msa;
 246     }
 247
 248     private void writeToFiles( final BasicSymmetricalDistanceMatrix m ) {
 249         if ( !ForesterUtil.isEmpty( _options.getIntermediateFilesBase() ) ) {
 250             try {
 251                 final BufferedWriter msa_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase()
 252                         + MSA_FILE_SUFFIX ) );
 253                 _msa.write( msa_writer );
 254                 msa_writer.close();
 255                 final BufferedWriter pwd_writer = new BufferedWriter( new FileWriter( _options.getIntermediateFilesBase()
 256                         + PWD_FILE_SUFFIX ) );
 257                 m.write( pwd_writer );
 258                 pwd_writer.close();
 259             }
 260             catch ( final Exception e ) {
 261                 System.out.println( "Error: " + e.getMessage() );
 262             }
 263         }
 264     }
 265
 266     public static void extractFastaInformation( final Phylogeny phy ) {
 267         for( final PhylogenyNodeIterator iter = phy.iteratorExternalForward(); iter.hasNext(); ) {
 268             final PhylogenyNode node = iter.next();
 269             if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 270                 final Matcher name_m = FastaParser.FASTA_DESC_LINE.matcher( node.getName() );
 271                 if ( name_m.lookingAt() ) {
 272                     System.out.println();
 273                     // System.out.println( name_m.group( 1 ) );
 274                     // System.out.println( name_m.group( 2 ) );
 275                     // System.out.println( name_m.group( 3 ) );
 276                     // System.out.println( name_m.group( 4 ) );
 277                     final String acc_source = name_m.group( 1 );
 278                     final String acc = name_m.group( 2 );
 279                     final String seq_name = name_m.group( 3 );
 280                     final String tax_sn = name_m.group( 4 );
 281                     if ( !ForesterUtil.isEmpty( acc_source ) && !ForesterUtil.isEmpty( acc ) ) {
 282                         AptxUtil.ensurePresenceOfSequence( node );
 283                         node.getNodeData().getSequence( 0 ).setAccession( new Accession( acc, acc_source ) );
 284                     }
 285                     if ( !ForesterUtil.isEmpty( seq_name ) ) {
 286                         AptxUtil.ensurePresenceOfSequence( node );
 287                         node.getNodeData().getSequence( 0 ).setName( seq_name );
 288                     }
 289                     if ( !ForesterUtil.isEmpty( tax_sn ) ) {
 290                         AptxUtil.ensurePresenceOfTaxonomy( node );
 291                         node.getNodeData().getTaxonomy( 0 ).setScientificName( tax_sn );
 292                     }
 293                 }
 294             }
 295         }
 296     }
 297 }