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         if ( gene_tree.getRoot().getNumberOfDescendants() == 3 ) {
  63             gene_tree.reRoot( gene_tree.getRoot().getChildNode( 2 ) );
  64         }
  65         final NodesLinkingResult nodes_linking_result = linkNodesOfG( gene_tree,
  66                                                                       species_tree,
  67                                                                       strip_gene_tree,
  68                                                                       strip_species_tree );
  69         _stripped_gene_tree_nodes = nodes_linking_result.getStrippedGeneTreeNodes();
  70         _stripped_species_tree_nodes = nodes_linking_result.getStrippedSpeciesTreeNodes();
  71         _mapped_species_tree_nodes = nodes_linking_result.getMappedSpeciesTreeNodes();
  72         _scientific_names_mapped_to_reduced_specificity = nodes_linking_result
  73                 .getScientificNamesMappedToReducedSpecificity();
  74         _tax_comp_base = nodes_linking_result.getTaxCompBase();
  75         PhylogenyMethods.preOrderReId( species_tree );
  76         final GSDIsummaryResult gsdi_summary_result = geneTreePostOrderTraversal( gene_tree,
  77                                                                                   _most_parsimonious_duplication_model );
  78         _speciation_or_duplication_events_sum = gsdi_summary_result.getSpeciationOrDuplicationEventsSum();
  79         _speciations_sum = gsdi_summary_result.getSpeciationsSum();
  80         _duplications_sum = gsdi_summary_result.getDuplicationsSum();
  81     }
  82
  83     public int getDuplicationsSum() {
  84         return _duplications_sum;
  85     }
  86
  87     @Override
  88     public Set<PhylogenyNode> getMappedExternalSpeciesTreeNodes() {
  89         return _mapped_species_tree_nodes;
  90     }
  91
  92     @Override
  93     public final SortedSet<String> getReMappedScientificNamesFromGeneTree() {
  94         return _scientific_names_mapped_to_reduced_specificity;
  95     }
  96
  97     public final int getSpeciationOrDuplicationEventsSum() {
  98         return _speciation_or_duplication_events_sum;
  99     }
 100
 101     @Override
 102     public final int getSpeciationsSum() {
 103         return _speciations_sum;
 104     }
 105
 106     @Override
 107     public List<PhylogenyNode> getStrippedExternalGeneTreeNodes() {
 108         return _stripped_gene_tree_nodes;
 109     }
 110
 111     @Override
 112     public List<PhylogenyNode> getStrippedSpeciesTreeNodes() {
 113         return _stripped_species_tree_nodes;
 114     }
 115
 116     @Override
 117     public TaxonomyComparisonBase getTaxCompBase() {
 118         return _tax_comp_base;
 119     }
 120
 121     @Override
 122     public final String toString() {
 123         final StringBuffer sb = new StringBuffer();
 124         sb.append( "Most parsimonious duplication model: " + _most_parsimonious_duplication_model );
 125         sb.append( ForesterUtil.getLineSeparator() );
 126         sb.append( "Speciations sum                    : " + getSpeciationsSum() );
 127         sb.append( ForesterUtil.getLineSeparator() );
 128         sb.append( "Duplications sum                   : " + getDuplicationsSum() );
 129         sb.append( ForesterUtil.getLineSeparator() );
 130         if ( !_most_parsimonious_duplication_model ) {
 131             sb.append( "Speciation or duplications sum     : " + getSpeciationOrDuplicationEventsSum() );
 132             sb.append( ForesterUtil.getLineSeparator() );
 133         }
 134         return sb.toString();
 135     }
 136
 137     /**
 138      * Traverses the subtree of PhylogenyNode g in postorder, calculating the
 139      * mapping function M, and determines which nodes represent speciation
 140      * events and which ones duplication events.
 141      * <p>
 142      * Preconditions: Mapping M for external nodes must have been calculated and
 143      * the species tree must be labeled in preorder.
 144      * <p>
 145      * @return
 146      * @throws SDIException
 147      *
 148      */
 149     final static GSDIsummaryResult geneTreePostOrderTraversal( final Phylogeny gene_tree,
 150                                                                final boolean most_parsimonious_duplication_model )
 151             throws SDIException {
 152         final GSDIsummaryResult res = new GSDIsummaryResult();
 153         for( final PhylogenyNodeIterator it = gene_tree.iteratorPostorder(); it.hasNext(); ) {
 154             final PhylogenyNode g = it.next();
 155             if ( g.isInternal() ) {
 156                 if ( g.getNumberOfDescendants() != 2 ) {
 157                     throw new SDIException( "gene tree contains internal node with " + g.getNumberOfDescendants()
 158                             + " descendents" );
 159                 }
 160                 PhylogenyNode s1 = g.getChildNode1().getLink();
 161                 PhylogenyNode s2 = g.getChildNode2().getLink();
 162                 while ( s1 != s2 ) {
 163                     if ( s1.getId() > s2.getId() ) {
 164                         s1 = s1.getParent();
 165                     }
 166                     else {
 167                         s2 = s2.getParent();
 168                     }
 169                 }
 170                 g.setLink( s1 );
 171                 determineEvent( s1, g, most_parsimonious_duplication_model, res );
 172             }
 173         }
 174         return res;
 175     }
 176
 177     final static NodesLinkingResult linkNodesOfG( final Phylogeny gene_tree,
 178                                                   final Phylogeny species_tree,
 179                                                   final boolean strip_gene_tree,
 180                                                   final boolean strip_species_tree ) throws SDIException {
 181         final TaxonomyComparisonBase tax_comp_base = SDIutil.determineTaxonomyComparisonBase( gene_tree );
 182         if ( tax_comp_base == null ) {
 183             throw new RuntimeException( "failed to establish taxonomy linking base (taxonomy linking base is null)" );
 184         }
 185         return linkNodesOfG( gene_tree, species_tree, tax_comp_base, strip_gene_tree, strip_species_tree );
 186     }
 187
 188     /**
 189      * This allows for linking of internal nodes of the species tree (as opposed
 190      * to just external nodes, as in the method it overrides.
 191      * If TaxonomyComparisonBase is null, it will try to determine it.
 192      * @throws SDIException
 193      *
 194      */
 195     final static NodesLinkingResult linkNodesOfG( final Phylogeny gene_tree,
 196                                                   final Phylogeny species_tree,
 197                                                   final TaxonomyComparisonBase tax_comp_base,
 198                                                   final boolean strip_gene_tree,
 199                                                   final boolean strip_species_tree ) throws SDIException {
 200         if ( tax_comp_base == null ) {
 201             throw new IllegalArgumentException( "taxonomy linking base is null" );
 202         }
 203         final Map<String, PhylogenyNode> species_to_node_map = new HashMap<String, PhylogenyNode>();
 204         final List<PhylogenyNode> species_tree_ext_nodes = new ArrayList<PhylogenyNode>();
 205         final NodesLinkingResult res = new NodesLinkingResult();
 206         res.setTaxCompBase( tax_comp_base );
 207         // Stringyfied taxonomy is the key, node is the value.
 208         for( final PhylogenyNodeIterator iter = species_tree.iteratorExternalForward(); iter.hasNext(); ) {
 209             final PhylogenyNode s = iter.next();
 210             species_tree_ext_nodes.add( s );
 211             if ( s.getNodeData().isHasTaxonomy() ) {
 212                 final String tax_str = SDIutil.taxonomyToString( s, res.getTaxCompBase() );
 213                 if ( !ForesterUtil.isEmpty( tax_str ) ) {
 214                     if ( species_to_node_map.containsKey( tax_str ) ) {
 215                         throw new SDIException( "taxonomy \"" + tax_str + "\" is not unique in species tree (using "
 216                                 + res.getTaxCompBase() + " for linking to gene tree)" );
 217                     }
 218                     species_to_node_map.put( tax_str, s );
 219                 }
 220             }
 221         }
 222         // Retrieve the reference to the node with a matching stringyfied taxonomy.
 223         for( final PhylogenyNodeIterator iter = gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
 224             final PhylogenyNode g = iter.next();
 225             if ( !g.getNodeData().isHasTaxonomy() ) {
 226                 if ( strip_gene_tree ) {
 227                     res.getStrippedGeneTreeNodes().add( g );
 228                 }
 229                 else {
 230                     throw new SDIException( "gene tree node \"" + g + "\" has no taxonomic data" );
 231                 }
 232             }
 233             else {
 234                 final String tax_str = SDIutil.taxonomyToString( g, res.getTaxCompBase() );
 235                 if ( ForesterUtil.isEmpty( tax_str ) ) {
 236                     if ( strip_gene_tree ) {
 237                         res.getStrippedGeneTreeNodes().add( g );
 238                     }
 239                     else {
 240                         throw new SDIException( "gene tree node \"" + g + "\" has no appropriate taxonomic data" );
 241                     }
 242                 }
 243                 else {
 244                     PhylogenyNode s = species_to_node_map.get( tax_str );
 245                     if ( ( res.getTaxCompBase() == TaxonomyComparisonBase.SCIENTIFIC_NAME ) && ( s == null )
 246                             && ( ForesterUtil.countChars( tax_str, ' ' ) > 1 ) ) {
 247                         s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 248                                                                 tax_str,
 249                                                                 res.getScientificNamesMappedToReducedSpecificity() );
 250                     }
 251                     if ( s == null ) {
 252                         if ( strip_gene_tree ) {
 253                             res.getStrippedGeneTreeNodes().add( g );
 254                         }
 255                         else {
 256                             throw new SDIException( "taxonomy \"" + g.getNodeData().getTaxonomy()
 257                                     + "\" not present in species tree" );
 258                         }
 259                     }
 260                     else {
 261                         g.setLink( s );
 262                         res.getMappedSpeciesTreeNodes().add( s );
 263                     }
 264                 }
 265             }
 266         } // for loop
 267         if ( strip_gene_tree ) {
 268             stripTree( gene_tree, res.getStrippedGeneTreeNodes() );
 269             if ( gene_tree.isEmpty() || ( gene_tree.getNumberOfExternalNodes() < 2 ) ) {
 270                 throw new SDIException( "species could not be mapped between gene tree and species tree (based on "
 271                         + res.getTaxCompBase() + ")" );
 272             }
 273         }
 274         if ( strip_species_tree ) {
 275             stripSpeciesTree( species_tree, species_tree_ext_nodes, res );
 276         }
 277         return res;
 278     }
 279
 280     private final static void addScientificNamesMappedToReducedSpecificity( final String s1,
 281                                                                             final String s2,
 282                                                                             final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 283         scientific_names_mapped_to_reduced_specificity.add( s1 + " -> " + s2 );
 284     }
 285
 286     private final static void determineEvent( final PhylogenyNode s,
 287                                               final PhylogenyNode g,
 288                                               final boolean most_parsimonious_duplication_model,
 289                                               final GSDIsummaryResult res ) {
 290         boolean oyako = false;
 291         if ( ( g.getChildNode1().getLink() == s ) || ( g.getChildNode2().getLink() == s ) ) {
 292             oyako = true;
 293         }
 294         if ( g.getLink().getNumberOfDescendants() == 2 ) {
 295             if ( oyako ) {
 296                 g.getNodeData().setEvent( Event.createSingleDuplicationEvent() );
 297                 res.increaseDuplicationsSum();
 298             }
 299             else {
 300                 g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 301                 res.increaseSpeciationsSum();
 302             }
 303         }
 304         else {
 305             if ( oyako ) {
 306                 final Set<PhylogenyNode> set = new HashSet<PhylogenyNode>();
 307                 for( PhylogenyNode n : g.getChildNode1().getAllExternalDescendants() ) {
 308                     n = n.getLink();
 309                     while ( n.getParent() != s ) {
 310                         n = n.getParent();
 311                         if ( n.isRoot() ) {
 312                             break;
 313                         }
 314                     }
 315                     set.add( n );
 316                 }
 317                 boolean multiple = false;
 318                 for( PhylogenyNode n : g.getChildNode2().getAllExternalDescendants() ) {
 319                     n = n.getLink();
 320                     while ( n.getParent() != s ) {
 321                         n = n.getParent();
 322                         if ( n.isRoot() ) {
 323                             break;
 324                         }
 325                     }
 326                     if ( set.contains( n ) ) {
 327                         multiple = true;
 328                         break;
 329                     }
 330                 }
 331                 if ( multiple ) {
 332                     g.getNodeData().setEvent( Event.createSingleDuplicationEvent() );
 333                     res.increaseDuplicationsSum();
 334                 }
 335                 else {
 336                     if ( most_parsimonious_duplication_model ) {
 337                         g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 338                         res.increaseSpeciationsSum();
 339                     }
 340                     else {
 341                         g.getNodeData().setEvent( Event.createSingleSpeciationOrDuplicationEvent() );
 342                         res.increaseSpeciationOrDuplicationEventsSum();
 343                     }
 344                 }
 345             }
 346             else {
 347                 g.getNodeData().setEvent( Event.createSingleSpeciationEvent() );
 348                 res.increaseSpeciationsSum();
 349             }
 350         }
 351     }
 352
 353     private final static void stripSpeciesTree( final Phylogeny species_tree,
 354                                                 final List<PhylogenyNode> species_tree_ext_nodes,
 355                                                 final NodesLinkingResult res ) {
 356         for( final PhylogenyNode s : species_tree_ext_nodes ) {
 357             if ( !res.getMappedSpeciesTreeNodes().contains( s ) ) {
 358                 species_tree.deleteSubtree( s, true );
 359                 res.getStrippedSpeciesTreeNodes().add( s );
 360             }
 361         }
 362         species_tree.clearHashIdToNodeMap();
 363         species_tree.externalNodesHaveChanged();
 364     }
 365
 366     private final static void stripTree( final Phylogeny phy, final List<PhylogenyNode> strip_nodes ) {
 367         for( final PhylogenyNode g : strip_nodes ) {
 368             phy.deleteSubtree( g, true );
 369         }
 370         phy.clearHashIdToNodeMap();
 371         phy.externalNodesHaveChanged();
 372     }
 373
 374     private final static PhylogenyNode tryMapByRemovingOverlySpecificData( final Map<String, PhylogenyNode> species_to_node_map,
 375                                                                            final String tax_str,
 376                                                                            final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 377         PhylogenyNode s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 378                                                               tax_str,
 379                                                               " (",
 380                                                               scientific_names_mapped_to_reduced_specificity );
 381         if ( s == null ) {
 382             if ( ForesterUtil.countChars( tax_str, ' ' ) == 2 ) {
 383                 final String new_tax_str = tax_str.substring( 0, tax_str.lastIndexOf( ' ' ) ).trim();
 384                 s = species_to_node_map.get( new_tax_str );
 385                 if ( s != null ) {
 386                     addScientificNamesMappedToReducedSpecificity( tax_str,
 387                                                                   new_tax_str,
 388                                                                   scientific_names_mapped_to_reduced_specificity );
 389                 }
 390             }
 391         }
 392         if ( s == null ) {
 393             for( final String t : new String[] { " subspecies ", " strain ", " variety ", " varietas ", " subvariety ",
 394                     " form ", " subform ", " cultivar ", " section ", " subsection " } ) {
 395                 s = tryMapByRemovingOverlySpecificData( species_to_node_map,
 396                                                         tax_str,
 397                                                         t,
 398                                                         scientific_names_mapped_to_reduced_specificity );
 399                 if ( s != null ) {
 400                     break;
 401                 }
 402             }
 403         }
 404         return s;
 405     }
 406
 407     private final static PhylogenyNode tryMapByRemovingOverlySpecificData( final Map<String, PhylogenyNode> species_to_node_map,
 408                                                                            final String tax_str,
 409                                                                            final String term,
 410                                                                            final SortedSet<String> scientific_names_mapped_to_reduced_specificity ) {
 411         final int i = tax_str.indexOf( term );
 412         if ( i > 4 ) {
 413             final String new_tax_str = tax_str.substring( 0, i ).trim();
 414             final PhylogenyNode s = species_to_node_map.get( new_tax_str );
 415             if ( s != null ) {
 416                 addScientificNamesMappedToReducedSpecificity( tax_str,
 417                                                               new_tax_str,
 418                                                               scientific_names_mapped_to_reduced_specificity );
 419             }
 420             return s;
 421         }
 422         return null;
 423     }
 424 }