forester/java/src/org/forester/sdi/GSDI.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2009 Christian M. Zmasek
   6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   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.sdi;
  27
  28 import java.util.ArrayList;
  29 import java.util.HashMap;
  30 import java.util.HashSet;
  31 import java.util.List;
  32 import java.util.Map;
  33 import java.util.Set;
  34 import java.util.SortedSet;
  35
  36 import org.forester.phylogeny.Phylogeny;
  37 import org.forester.phylogeny.PhylogenyMethods;
  38 import org.forester.phylogeny.PhylogenyNode;
  39 import org.forester.phylogeny.data.Event;
  40 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  41 import org.forester.sdi.SDIutil.TaxonomyComparisonBase;
  42 import org.forester.util.ForesterUtil;
  43
  44 public final class GSDI implements GSDII  {
  45
  46     private final boolean                _most_parsimonious_duplication_model;
  47     private final int                    _speciation_or_duplication_events_sum;
  48     private final int                    _speciations_sum;
  49     private final int                    _duplications_sum;
  50     private final List<PhylogenyNode>    _stripped_gene_tree_nodes;
  51     private final List<PhylogenyNode>    _stripped_species_tree_nodes;
  52     private final Set<PhylogenyNode>     _mapped_species_tree_nodes;
  53     private final TaxonomyComparisonBase _tax_comp_base;
  54     private final SortedSet<String>      _scientific_names_mapped_to_reduced_specificity;
  55
  56     public GSDI( final Phylogeny gene_tree,
  57                  final Phylogeny species_tree,
  58                  final boolean most_parsimonious_duplication_model,
  59                  final boolean strip_gene_tree,
  60                  final boolean strip_species_tree ) throws SDIException {
  61         _most_parsimonious_duplication_model = most_parsimonious_duplication_model;
  62         final NodesLinkingResult nodes_linking_result = linkNodesOfG( gene_tree,
  63                                                                       species_tree,
  64                                                                       null,
  65                                                                       strip_gene_tree,
  66                                                                       strip_species_tree );
  67         _stripped_gene_tree_nodes = nodes_linking_result.getStrippedGeneTreeNodes();
  68         _stripped_species_tree_nodes = nodes_linking_result.getStrippedSpeciesTreeNodes();
  69         _mapped_species_tree_nodes = nodes_linking_result.getMappedSpeciesTreeNodes();
  70         _scientific_names_mapped_to_reduced_specificity = nodes_linking_result
  71                 .getScientificNamesMappedToReducedSpecificity();
  72         _tax_comp_base = nodes_linking_result.getTaxCompBase();
  73         PhylogenyMethods.preOrderReId( species_tree );
  74         final GSDIsummaryResult gsdi_summary_result = geneTreePostOrderTraversal( gene_tree,
  75                                                                                   _most_parsimonious_duplication_model );
  76         _speciation_or_duplication_events_sum = gsdi_summary_result.getSpeciationOrDuplicationEventsSum();
  77         _speciations_sum = gsdi_summary_result.getSpeciationsSum();
  78         _duplications_sum = gsdi_summary_result.getDuplicationsSum();
  79     }
  80
  81     public int getDuplicationsSum() {
  82         return _duplications_sum;
  83     }
  84
  85     public Set<PhylogenyNode> getMappedExternalSpeciesTreeNodes() {
  86         return _mapped_species_tree_nodes;
  87     }
  88
  89     public final SortedSet<String> getReMappedScientificNamesFromGeneTree() {
  90         return _scientific_names_mapped_to_reduced_specificity;
  91     }
  92
  93     public final int getSpeciationOrDuplicationEventsSum() {
  94         return _speciation_or_duplication_events_sum;
  95     }
  96
  97     public final int getSpeciationsSum() {
  98         return _speciations_sum;
  99     }
 100
 101     public List<PhylogenyNode> getStrippedExternalGeneTreeNodes() {
 102         return _stripped_gene_tree_nodes;
 103     }
 104
 105     public List<PhylogenyNode> getStrippedSpeciesTreeNodes() {
 106         return _stripped_species_tree_nodes;
 107     }
 108
 109     public TaxonomyComparisonBase getTaxCompBase() {
 110         return _tax_comp_base;
 111     }
 112
 113     @Override
 114     public final String toString() {
 115         final StringBuffer sb = new StringBuffer();
 116         sb.append( "Most parsimonious duplication model: " + _most_parsimonious_duplication_model );
 117         sb.append( ForesterUtil.getLineSeparator() );
 118         sb.append( "Speciations sum                    : " + getSpeciationsSum() );
 119         sb.append( ForesterUtil.getLineSeparator() );
 120         sb.append( "Duplications sum                   : " + getDuplicationsSum() );
 121         sb.append( ForesterUtil.getLineSeparator() );
 122         if ( !_most_parsimonious_duplication_model ) {
 123             sb.append( "Speciation or duplications sum     : " + getSpeciationOrDuplicationEventsSum() );
 124             sb.append( ForesterUtil.getLineSeparator() );
 125         }
 126         return sb.toString();
 127     }
 128
 129     /**
 130      * Traverses the subtree of PhylogenyNode g in postorder, calculating the
 131      * mapping function M, and determines which nodes represent speciation
 132      * events and which ones duplication events.
 133      * <p>
 134      * Preconditions: Mapping M for external nodes must have been calculated and
 135      * the species tree must be labeled in preorder.
 136      * <p>
 137      * @return
 138      *
 139      */
 140     final static GSDIsummaryResult geneTreePostOrderTraversal( final Phylogeny gene_tree,
 141                                                                final boolean most_parsimonious_duplication_model ) {
 142         final GSDIsummaryResult res = new GSDIsummaryResult();
 143         for( final PhylogenyNodeIterator it = gene_tree.iteratorPostorder(); it.hasNext(); ) {
 144             final PhylogenyNode g = it.next();
 145             if ( g.isInternal() ) {
 146                 PhylogenyNode s1 = g.getChildNode1().getLink();
 147                 PhylogenyNode s2 = g.getChildNode2().getLink();
 148                 while ( s1 != s2 ) {
 149                     if ( s1.getId() > s2.getId() ) {
 150                         s1 = s1.getParent();
 151                     }
 152                     else {
 153                         s2 = s2.getParent();
 154                     }
 155                 }
 156                 g.setLink( s1 );
 157                 determineEvent( s1, g, most_parsimonious_duplication_model, res );
 158             }
 159         }
 160         return res;
 161     }
 162
 163     /**
 164      * This allows for linking of internal nodes of the species tree (as opposed
 165      * to just external nodes, as in the method it overrides.
 166      * If TaxonomyComparisonBase is null, it will try to determine it.
 167      * @throws SDIException
 168      *
 169      */
 170     final static NodesLinkingResult linkNodesOfG( final Phylogeny gene_tree,
 171                                                   final Phylogeny species_tree,
 172                                                   final TaxonomyComparisonBase tax_comp_base,
 173                                                   final boolean strip_gene_tree,
 174                                                   final boolean strip_species_tree ) throws SDIException {
 175         final Map<String, PhylogenyNode> species_to_node_map = new HashMap<String, PhylogenyNode>();
 176         final List<PhylogenyNode> species_tree_ext_nodes = new ArrayList<PhylogenyNode>();
 177         final NodesLinkingResult res = new NodesLinkingResult();
 178         if ( tax_comp_base == null ) {
 179             res.setTaxCompBase( SDIutil.determineTaxonomyComparisonBase( gene_tree ) );
 180         }
 181         else {
 182             res.setTaxCompBase( tax_comp_base );
 183         }
 184         // Stringyfied taxonomy is the key, node is the value.
 185         for( final PhylogenyNodeIterator iter = species_tree.iteratorExternalForward(); iter.hasNext(); ) {
 186             final PhylogenyNode s = iter.next();
 187             species_tree_ext_nodes.add( s );
 188             if ( s.getNodeData().isHasTaxonomy() ) {
 189                 final String tax_str = SDIutil.taxonomyToString( s, res.getTaxCompBase() );
 190                 if ( !ForesterUtil.isEmpty( tax_str ) ) {
 191                     if ( species_to_node_map.containsKey( tax_str ) ) {
 192                         throw new SDIException( "taxonomy \"" + s + "\" is not unique in species tree" );
 193                     }
 194                     species_to_node_map.put( tax_str, s );
 195                 }
 196             }
 197         }
 198         // Retrieve the reference to the node with a matching stringyfied taxonomy.
 199         for( final PhylogenyNodeIterator iter = gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
 200             final PhylogenyNode g = iter.next();
 201             if ( !g.getNodeData().isHasTaxonomy() ) {
 202                 if ( strip_gene_tree ) {
 203                     res.getStrippedGeneTreeNodes().add( g );
 204                 }
 205                 else {
 206                     throw new SDIException( "gene tree node \"" + g + "\" has no taxonomic data" );
 207                 }
 208             }
 209             else {
 210                 final String tax_str = SDIutil.taxonomyToString( g, res.getTaxCompBase() );
 211                 if ( ForesterUtil.isEmpty( tax_str ) ) {
 212                     if ( strip_gene_tree ) {
 213                         res.getStrippedGeneTreeNodes().add( g );
 214                     }
 215                     else {
 216                         throw new SDIException( "gene tree node \"" + g + "\" has no appropriate taxonomic data" );
 217                     }
 218                 }
 219                 else {
 220                     PhylogenyNode s = species_to_node_map.get( tax_str );
 221                     if ( ( res.getTaxCompBase() == TaxonomyComparisonBase.SCIENTIFIC_NAME ) && ( s == null )
 222                             && ( ForesterUtil.countChars( tax_str, ' ' ) > 1 ) ) {
 223                         s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 224                                                                 tax_str,
 225                                                                 res.getScientificNamesMappedToReducedSpecificity() );
 226                     }
 227                     if ( s == null ) {
 228                         if ( strip_gene_tree ) {
 229                             res.getStrippedGeneTreeNodes().add( g );
 230                         }
 231                         else {
 232                             throw new SDIException( "taxonomy \"" + g.getNodeData().getTaxonomy()
 233                                     + "\" not present in species tree" );
 234                         }
 235                     }
 236                     else {
 237                         g.setLink( s );
 238                         res.getMappedSpeciesTreeNodes().add( s );
 239                     }
 240                 }
 241             }
 242         } // for loop
 243         if ( strip_gene_tree ) {
 244             stripTree( gene_tree, res.getStrippedGeneTreeNodes() );
 245             if ( gene_tree.isEmpty() || ( gene_tree.getNumberOfExternalNodes() < 2 ) ) {
 246                 throw new SDIException( "species could not be mapped between gene tree and species tree" );
 247             }
 248         }
 249         if ( strip_species_tree ) {
 250             stripSpeciesTree( species_tree, species_tree_ext_nodes, res );
 251         }
 252         return res;
 253     }
 254
 255     private final static void addScientificNamesMappedToReducedSpecificity( final String s1,
 256                                                                             final String s2,
 257                                                                             final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 258         scientific_names_mapped_to_reduced_specificity.add( s1 + " -> " + s2 );
 259     }
 260
 261     private final static void determineEvent( final PhylogenyNode s,
 262                                               final PhylogenyNode g,
 263                                               final boolean most_parsimonious_duplication_model,
 264                                               final GSDIsummaryResult res ) {
 265         boolean oyako = false;
 266         if ( ( g.getChildNode1().getLink() == s ) || ( g.getChildNode2().getLink() == s ) ) {
 267             oyako = true;
 268         }
 269         if ( g.getLink().getNumberOfDescendants() == 2 ) {
 270             if ( oyako ) {
 271                 g.getNodeData().setEvent( Event.createSingleDuplicationEvent() );
 272                 res.increaseDuplicationsSum();
 273             }
 274             else {
 275                 g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 276                 res.increaseSpeciationsSum();
 277             }
 278         }
 279         else {
 280             if ( oyako ) {
 281                 final Set<PhylogenyNode> set = new HashSet<PhylogenyNode>();
 282                 for( PhylogenyNode n : g.getChildNode1().getAllExternalDescendants() ) {
 283                     n = n.getLink();
 284                     while ( n.getParent() != s ) {
 285                         n = n.getParent();
 286                         if ( n.isRoot() ) {
 287                             break;
 288                         }
 289                     }
 290                     set.add( n );
 291                 }
 292                 boolean multiple = false;
 293                 for( PhylogenyNode n : g.getChildNode2().getAllExternalDescendants() ) {
 294                     n = n.getLink();
 295                     while ( n.getParent() != s ) {
 296                         n = n.getParent();
 297                         if ( n.isRoot() ) {
 298                             break;
 299                         }
 300                     }
 301                     if ( set.contains( n ) ) {
 302                         multiple = true;
 303                         break;
 304                     }
 305                 }
 306                 if ( multiple ) {
 307                     g.getNodeData().setEvent( Event.createSingleDuplicationEvent() );
 308                     res.increaseDuplicationsSum();
 309                 }
 310                 else {
 311                     if ( most_parsimonious_duplication_model ) {
 312                         g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 313                         res.increaseSpeciationsSum();
 314                     }
 315                     else {
 316                         g.getNodeData().setEvent( Event.createSingleSpeciationOrDuplicationEvent() );
 317                         res.increaseSpeciationOrDuplicationEventsSum();
 318                     }
 319                 }
 320             }
 321             else {
 322                 g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 323                 res.increaseSpeciationsSum();
 324             }
 325         }
 326     }
 327
 328     private final static void stripSpeciesTree( final Phylogeny species_tree,
 329                                                 final List<PhylogenyNode> species_tree_ext_nodes,
 330                                                 final NodesLinkingResult res ) {
 331         for( final PhylogenyNode s : species_tree_ext_nodes ) {
 332             if ( !res.getMappedSpeciesTreeNodes().contains( s ) ) {
 333                 species_tree.deleteSubtree( s, true );
 334                 res.getStrippedSpeciesTreeNodes().add( s );
 335             }
 336         }
 337         species_tree.clearHashIdToNodeMap();
 338         species_tree.externalNodesHaveChanged();
 339     }
 340
 341     private final static void stripTree( final Phylogeny phy, final List<PhylogenyNode> strip_nodes ) {
 342         for( final PhylogenyNode g : strip_nodes ) {
 343             phy.deleteSubtree( g, true );
 344         }
 345         phy.clearHashIdToNodeMap();
 346         phy.externalNodesHaveChanged();
 347     }
 348
 349     private final static PhylogenyNode tryMapByRemovingOverlySpecificData( final Map<String, PhylogenyNode> species_to_node_map,
 350                                                                            final String tax_str,
 351                                                                            final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 352         PhylogenyNode s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 353                                                               tax_str,
 354                                                               " (",
 355                                                               scientific_names_mapped_to_reduced_specificity );
 356         if ( s == null ) {
 357             if ( ForesterUtil.countChars( tax_str, ' ' ) == 2 ) {
 358                 final String new_tax_str = tax_str.substring( 0, tax_str.lastIndexOf( ' ' ) ).trim();
 359                 s = species_to_node_map.get( new_tax_str );
 360                 if ( s != null ) {
 361                     addScientificNamesMappedToReducedSpecificity( tax_str,
 362                                                                   new_tax_str,
 363                                                                   scientific_names_mapped_to_reduced_specificity );
 364                 }
 365             }
 366         }
 367         if ( s == null ) {
 368             for( final String t : new String[] { " subspecies ", " strain ", " variety ", " varietas ", " subvariety ",
 369                     " form ", " subform ", " cultivar ", " section ", " subsection " } ) {
 370                 s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 371                                                         tax_str,
 372                                                         t,
 373                                                         scientific_names_mapped_to_reduced_specificity );
 374                 if ( s != null ) {
 375                     break;
 376                 }
 377             }
 378         }
 379         return s;
 380     }
 381
 382     private final static PhylogenyNode tryMapByRemovingOverlySpecificData( final Map<String, PhylogenyNode> species_to_node_map,
 383                                                                            final String tax_str,
 384                                                                            final String term,
 385                                                                            final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 386         final int i = tax_str.indexOf( term );
 387         if ( i > 4 ) {
 388             final String new_tax_str = tax_str.substring( 0, i ).trim();
 389             final PhylogenyNode s = species_to_node_map.get( new_tax_str );
 390             if ( s != null ) {
 391                 addScientificNamesMappedToReducedSpecificity( tax_str,
 392                                                               new_tax_str,
 393                                                               scientific_names_mapped_to_reduced_specificity );
 394             }
 395             return s;
 396         }
 397         return null;
 398     }
 399 }