2 // FORESTER -- software libraries and applications
3 // for evolutionary biology research and applications.
5 // Copyright (C) 2008-2009 Christian M. Zmasek
6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
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.
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.
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
23 // Contact: phylosoft @ gmail . com
24 // WWW: www.phylosoft.org/forester
26 package org.forester.sdi;
28 import java.util.ArrayList;
29 import java.util.HashMap;
30 import java.util.HashSet;
31 import java.util.List;
35 import org.forester.phylogeny.Phylogeny;
36 import org.forester.phylogeny.PhylogenyNode;
37 import org.forester.phylogeny.data.Event;
38 import org.forester.phylogeny.data.Taxonomy;
39 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
40 import org.forester.util.ForesterUtil;
43 * Implements our algorithm for speciation - duplication inference (SDI). <p>
44 * The initialization is accomplished by: </p> <ul> <li>method
45 * "linkExtNodesOfG()" of class SDI: setting the links for the external nodes of
46 * the gene tree <li>"preorderReID(int)" from class Phylogeny: numbering of
47 * nodes of the species tree in preorder <li>the optional stripping of the
48 * species tree is accomplished by method "stripTree(Phylogeny,Phylogeny)" of
49 * class Phylogeny </ul> <p> The recursion part is accomplished by this class'
50 * method "geneTreePostOrderTraversal(PhylogenyNode)". <p> Requires JDK 1.5 or
53 * @see SDI#linkNodesOfG()
55 * @see Phylogeny#preorderReID(int)
58 * PhylogenyMethods#taxonomyBasedDeletionOfExternalNodes(Phylogeny,Phylogeny)
60 * @see #geneTreePostOrderTraversal(PhylogenyNode)
62 * @author Christian M. Zmasek
64 public final class GSDI extends SDI {
66 private final HashMap<PhylogenyNode, Integer> _transversal_counts;
67 private final boolean _most_parsimonious_duplication_model;
68 private final boolean _strip_gene_tree;
69 private final boolean _strip_species_tree;
70 private int _speciation_or_duplication_events_sum;
71 private int _speciations_sum;
72 private final List<PhylogenyNode> _stripped_gene_tree_nodes;
73 private final List<PhylogenyNode> _stripped_species_tree_nodes;
74 private final Set<PhylogenyNode> _mapped_species_tree_nodes;
77 * Constructor which sets the gene tree and the species tree to be compared.
78 * species_tree is the species tree to which the gene tree gene_tree will be
79 * compared to - with method "infer(boolean)". Both Trees must be completely
80 * binary and rooted. The actual inference is accomplished with method
81 * "infer(boolean)". The mapping cost L can then be calculated with method
82 * "computeMappingCost()".
85 * @see #infer(boolean)
86 * @see SDI#computeMappingCostL()
88 * reference to a rooted gene tree to which assign duplication vs
89 * speciation, must have species names in the species name fields
90 * for all external nodes
92 * reference to a rooted binary species tree which might get
93 * stripped in the process, must have species names in the
94 * species name fields for all external nodes
96 * @param most_parsimonious_duplication_model
97 * set to true to assign nodes as speciations which would
98 * otherwise be assiged as unknown because of polytomies in the
100 * @throws SdiException
103 public GSDI( final Phylogeny gene_tree,
104 final Phylogeny species_tree,
105 final boolean most_parsimonious_duplication_model,
106 final boolean strip_gene_tree,
107 final boolean strip_species_tree ) throws SdiException {
108 super( gene_tree, species_tree );
109 _speciation_or_duplication_events_sum = 0;
110 _speciations_sum = 0;
111 _most_parsimonious_duplication_model = most_parsimonious_duplication_model;
112 _transversal_counts = new HashMap<PhylogenyNode, Integer>();
113 _duplications_sum = 0;
114 _strip_gene_tree = strip_gene_tree;
115 _strip_species_tree = strip_species_tree;
116 _stripped_gene_tree_nodes = new ArrayList<PhylogenyNode>();
117 _stripped_species_tree_nodes = new ArrayList<PhylogenyNode>();
118 _mapped_species_tree_nodes = new HashSet<PhylogenyNode>();
119 getSpeciesTree().preOrderReId();
121 geneTreePostOrderTraversal( getGeneTree().getRoot() );
124 GSDI( final Phylogeny gene_tree, final Phylogeny species_tree, final boolean most_parsimonious_duplication_model )
125 throws SdiException {
126 this( gene_tree, species_tree, most_parsimonious_duplication_model, false, false );
129 private final Event createDuplicationEvent() {
130 final Event event = Event.createSingleDuplicationEvent();
135 private final Event createSingleSpeciationOrDuplicationEvent() {
136 final Event event = Event.createSingleSpeciationOrDuplicationEvent();
137 ++_speciation_or_duplication_events_sum;
141 private final Event createSpeciationEvent() {
142 final Event event = Event.createSingleSpeciationEvent();
147 // s is the node on the species tree g maps to.
148 private final void determineEvent( final PhylogenyNode s, final PhylogenyNode g ) {
150 // Determine how many children map to same node as parent.
151 int sum_g_childs_mapping_to_s = 0;
152 for( int i = 0; i < g.getNumberOfDescendants(); ++i ) {
153 final PhylogenyNode c = g.getChildNode( i );
154 if ( c.getLink() == s ) {
155 ++sum_g_childs_mapping_to_s;
158 // Determine the sum of traversals.
159 int traversals_sum = 0;
160 int max_traversals = 0;
161 PhylogenyNode max_traversals_node = null;
162 if ( !s.isExternal() ) {
163 for( int i = 0; i < s.getNumberOfDescendants(); ++i ) {
164 final PhylogenyNode current_node = s.getChildNode( i );
165 final int traversals = getTraversalCount( current_node );
166 traversals_sum += traversals;
167 if ( traversals > max_traversals ) {
168 max_traversals = traversals;
169 max_traversals_node = current_node;
173 // System.out.println( " sum=" + traversals_sum );
174 // System.out.println( " max=" + max_traversals );
175 // System.out.println( " m=" + sum_g_childs_mapping_to_s );
176 if ( sum_g_childs_mapping_to_s > 0 ) {
177 if ( traversals_sum == 2 ) {
178 event = createDuplicationEvent();
179 System.out.print( g.toString() );
180 System.out.println( " : ==2" );
181 // _transversal_counts.clear();
183 else if ( traversals_sum > 2 ) {
184 if ( max_traversals <= 1 ) {
185 if ( _most_parsimonious_duplication_model ) {
186 event = createSpeciationEvent();
189 event = createSingleSpeciationOrDuplicationEvent();
193 event = createDuplicationEvent();
194 //System.out.println( g.toString() );
195 _transversal_counts.put( max_traversals_node, 1 );
196 // _transversal_counts.clear();
200 event = createDuplicationEvent();
201 // _transversal_counts.clear();
203 normalizeTcounts( s );
206 event = createSpeciationEvent();
208 g.getNodeData().setEvent( event );
211 private void normalizeTcounts( final PhylogenyNode s ) {
212 int min_traversals = Integer.MAX_VALUE;
213 for( int i = 0; i < s.getNumberOfDescendants(); ++i ) {
214 final PhylogenyNode current_node = s.getChildNode( i );
215 final int traversals = getTraversalCount( current_node );
216 if ( traversals < min_traversals ) {
217 min_traversals = traversals;
220 for( int i = 0; i < s.getNumberOfDescendants(); ++i ) {
221 final PhylogenyNode current_node = s.getChildNode( i );
222 _transversal_counts.put( current_node, getTraversalCount( current_node ) - min_traversals );
227 * Traverses the subtree of PhylogenyNode g in postorder, calculating the
228 * mapping function M, and determines which nodes represent speciation
229 * events and which ones duplication events.
231 * Preconditions: Mapping M for external nodes must have been calculated and
232 * the species tree must be labeled in preorder.
236 * starting node of a gene tree - normally the root
238 final void geneTreePostOrderTraversal( final PhylogenyNode g ) {
239 if ( !g.isExternal() ) {
240 boolean all_ext = true;
241 for( int i = 0; i < g.getNumberOfDescendants(); ++i ) {
242 if ( g.getChildNode( i ).isInternal() ) {
248 //_transversal_counts.clear();
250 for( int i = 0; i < g.getNumberOfDescendants(); ++i ) {
251 geneTreePostOrderTraversal( g.getChildNode( i ) );
253 final PhylogenyNode[] linked_nodes = new PhylogenyNode[ g.getNumberOfDescendants() ];
254 for( int i = 0; i < linked_nodes.length; ++i ) {
255 if ( g.getChildNode( i ).getLink() == null ) {
256 System.out.println( "link is null for " + g.getChildNode( i ) );
259 linked_nodes[ i ] = g.getChildNode( i ).getLink();
261 final int[] min_max = obtainMinMaxIdIndices( linked_nodes );
262 int min_i = min_max[ 0 ];
263 int max_i = min_max[ 1 ];
264 // initTransversalCounts();
265 while ( linked_nodes[ min_i ] != linked_nodes[ max_i ] ) {
266 increaseTraversalCount( linked_nodes[ max_i ] );
267 linked_nodes[ max_i ] = linked_nodes[ max_i ].getParent();
268 final int[] min_max_ = obtainMinMaxIdIndices( linked_nodes );
269 min_i = min_max_[ 0 ];
270 max_i = min_max_[ 1 ];
272 final PhylogenyNode s = linked_nodes[ max_i ];
274 // Determines whether dup. or spec.
275 determineEvent( s, g );
279 public final int getSpeciationOrDuplicationEventsSum() {
280 return _speciation_or_duplication_events_sum;
283 public final int getSpeciationsSum() {
284 return _speciations_sum;
287 private final int getTraversalCount( final PhylogenyNode node ) {
288 if ( _transversal_counts.containsKey( node ) ) {
289 return _transversal_counts.get( node );
294 private final void increaseTraversalCount( final PhylogenyNode node ) {
295 if ( _transversal_counts.containsKey( node ) ) {
296 _transversal_counts.put( node, _transversal_counts.get( node ) + 1 );
299 _transversal_counts.put( node, 1 );
301 // System.out.println( "count for node " + node.getID() + " is now "
302 // + getTraversalCount( node ) );
306 * This allows for linking of internal nodes of the species tree (as opposed
307 * to just external nodes, as in the method it overrides.
308 * @throws SdiException
312 // final void linkNodesOfG() {
313 // final HashMap<Taxonomy, PhylogenyNode> speciestree_ext_nodes = createTaxonomyToNodeMap();
314 // if ( _strip_gene_tree ) {
315 // stripGeneTree( speciestree_ext_nodes );
316 // if ( ( _gene_tree == null ) || ( _gene_tree.getNumberOfExternalNodes() < 2 ) ) {
317 // throw new IllegalArgumentException( "species tree does not contain any"
318 // + " nodes matching species in the gene tree" );
321 // // Retrieve the reference to the PhylogenyNode with a matching species.
322 // for( final PhylogenyNodeIterator iter = _gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
323 // final PhylogenyNode g = iter.next();
324 // if ( !g.getNodeData().isHasTaxonomy() ) {
325 // throw new IllegalArgumentException( "gene tree node " + g + " has no taxonomic data" );
327 // final PhylogenyNode s = speciestree_ext_nodes.get( g.getNodeData().getTaxonomy() );
328 // if ( s == null ) {
329 // throw new IllegalArgumentException( "species " + g.getNodeData().getTaxonomy()
330 // + " not present in species tree" );
335 final void linkNodesOfG() throws SdiException {
336 final Map<String, PhylogenyNode> species_to_node_map = new HashMap<String, PhylogenyNode>();
337 final List<PhylogenyNode> species_tree_ext_nodes = new ArrayList<PhylogenyNode>();
338 final TaxonomyComparisonBase tax_comp_base = determineTaxonomyComparisonBase( _gene_tree );
339 // System.out.println( "comp base is: " + tax_comp_base );
340 // Stringyfied taxonomy is the key, node is the value.
341 for( final PhylogenyNodeIterator iter = _species_tree.iteratorExternalForward(); iter.hasNext(); ) {
342 final PhylogenyNode s = iter.next();
343 species_tree_ext_nodes.add( s );
344 final String tax_str = taxonomyToString( s, tax_comp_base );
345 if ( !ForesterUtil.isEmpty( tax_str ) ) {
346 if ( species_to_node_map.containsKey( tax_str ) ) {
347 throw new SdiException( "taxonomy \"" + s + "\" is not unique in species tree" );
349 species_to_node_map.put( tax_str, s );
352 // Retrieve the reference to the node with a matching stringyfied taxonomy.
353 for( final PhylogenyNodeIterator iter = _gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
354 final PhylogenyNode g = iter.next();
355 if ( !g.getNodeData().isHasTaxonomy() ) {
356 if ( _strip_gene_tree ) {
357 _stripped_gene_tree_nodes.add( g );
360 throw new SdiException( "gene tree node \"" + g + "\" has no taxonomic data" );
364 final String tax_str = taxonomyToString( g, tax_comp_base );
365 if ( ForesterUtil.isEmpty( tax_str ) ) {
366 if ( _strip_gene_tree ) {
367 _stripped_gene_tree_nodes.add( g );
370 throw new SdiException( "gene tree node \"" + g + "\" has no appropriate taxonomic data" );
374 final PhylogenyNode s = species_to_node_map.get( tax_str );
376 if ( _strip_gene_tree ) {
377 _stripped_gene_tree_nodes.add( g );
380 throw new SdiException( "taxonomy \"" + g.getNodeData().getTaxonomy()
381 + "\" not present in species tree" );
386 _mapped_species_tree_nodes.add( s );
387 // System.out.println( "setting link of " + g + " to " + s );
392 if ( _strip_gene_tree ) {
393 for( final PhylogenyNode g : _stripped_gene_tree_nodes ) {
394 _gene_tree.deleteSubtree( g, true );
397 if ( _strip_species_tree ) {
398 for( final PhylogenyNode s : species_tree_ext_nodes ) {
399 if ( !_mapped_species_tree_nodes.contains( s ) ) {
400 _species_tree.deleteSubtree( s, true );
406 public Set<PhylogenyNode> getMappedExternalSpeciesTreeNodes() {
407 return _mapped_species_tree_nodes;
410 // final private HashMap<Taxonomy, PhylogenyNode> createTaxonomyToNodeMap() {
411 // final HashMap<Taxonomy, PhylogenyNode> speciestree_ext_nodes = new HashMap<Taxonomy, PhylogenyNode>();
412 // for( final PhylogenyNodeIterator iter = _species_tree.iteratorLevelOrder(); iter.hasNext(); ) {
413 // final PhylogenyNode n = iter.next();
414 // if ( n.getNodeData().isHasTaxonomy() ) {
415 // if ( speciestree_ext_nodes.containsKey( n.getNodeData().getTaxonomy() ) ) {
416 // throw new IllegalArgumentException( "taxonomy [" + n.getNodeData().getTaxonomy()
417 // + "] is not unique in species phylogeny" );
419 // speciestree_ext_nodes.put( n.getNodeData().getTaxonomy(), n );
422 // return speciestree_ext_nodes;
424 // private final void stripGeneTree( final HashMap<Taxonomy, PhylogenyNode> speciestree_ext_nodes ) {
425 // // final Set<PhylogenyNode> to_delete = new HashSet<PhylogenyNode>();
426 // for( final PhylogenyNodeIterator iter = _gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
427 // final PhylogenyNode g = iter.next();
428 // if ( !g.getNodeData().isHasTaxonomy() ) {
429 // throw new IllegalArgumentException( "gene tree node " + g + " has no taxonomic data" );
431 // if ( !speciestree_ext_nodes.containsKey( g.getNodeData().getTaxonomy() ) ) {
432 // _stripped_gene_tree_nodes.add( g );
435 // for( final PhylogenyNode n : _stripped_gene_tree_nodes ) {
436 // _gene_tree.deleteSubtree( n, true );
439 // private final void stripGeneTree2( final HashMap<Taxonomy, PhylogenyNode> speciestree_ext_nodes ) {
440 // // final Set<PhylogenyNode> to_delete = new HashSet<PhylogenyNode>();
441 // for( final PhylogenyNodeIterator iter = _gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
442 // final PhylogenyNode g = iter.next();
443 // if ( !g.getNodeData().isHasTaxonomy() ) {
444 // _stripped_gene_tree_nodes.add( g );
447 // if ( !speciestree_ext_nodes.containsKey( g.getNodeData().getTaxonomy() ) ) {
448 // _stripped_gene_tree_nodes.add( g );
452 // for( final PhylogenyNode n : _stripped_gene_tree_nodes ) {
453 // _gene_tree.deleteSubtree( n, true );
456 public static TaxonomyComparisonBase determineTaxonomyComparisonBase( final Phylogeny gene_tree ) {
457 int with_id_count = 0;
458 int with_code_count = 0;
459 int with_sn_count = 0;
461 for( final PhylogenyNodeIterator iter = gene_tree.iteratorExternalForward(); iter.hasNext(); ) {
462 final PhylogenyNode g = iter.next();
463 if ( g.getNodeData().isHasTaxonomy() ) {
464 final Taxonomy tax = g.getNodeData().getTaxonomy();
465 if ( ( tax.getIdentifier() != null ) && !ForesterUtil.isEmpty( tax.getIdentifier().getValue() ) ) {
466 if ( ++with_id_count > max ) {
470 if ( !ForesterUtil.isEmpty( tax.getTaxonomyCode() ) ) {
471 if ( ++with_code_count > max ) {
472 max = with_code_count;
475 if ( !ForesterUtil.isEmpty( tax.getScientificName() ) ) {
476 if ( ++with_sn_count > max ) {
483 throw new IllegalArgumentException( "gene tree has no taxonomic data" );
485 else if ( max == 1 ) {
486 throw new IllegalArgumentException( "gene tree has only one node with taxonomic data" );
488 else if ( max == with_sn_count ) {
489 return SDI.TaxonomyComparisonBase.SCIENTIFIC_NAME;
491 else if ( max == with_id_count ) {
492 return SDI.TaxonomyComparisonBase.ID;
495 return SDI.TaxonomyComparisonBase.CODE;
499 public List<PhylogenyNode> getStrippedExternalGeneTreeNodes() {
500 return _stripped_gene_tree_nodes;
504 public final String toString() {
505 final StringBuffer sb = new StringBuffer();
506 sb.append( "Most parsimonious duplication model: " + _most_parsimonious_duplication_model );
507 sb.append( ForesterUtil.getLineSeparator() );
508 sb.append( "Speciations sum : " + getSpeciationsSum() );
509 sb.append( ForesterUtil.getLineSeparator() );
510 sb.append( "Duplications sum : " + getDuplicationsSum() );
511 sb.append( ForesterUtil.getLineSeparator() );
512 if ( !_most_parsimonious_duplication_model ) {
513 sb.append( "Speciation or duplications sum : " + getSpeciationOrDuplicationEventsSum() );
514 sb.append( ForesterUtil.getLineSeparator() );
516 sb.append( "mapping cost L : " + computeMappingCostL() );
517 return sb.toString();
520 static final int[] obtainMinMaxIdIndices( final PhylogenyNode[] linked_nodes ) {
523 int max_i_id = -Integer.MAX_VALUE;
524 int min_i_id = Integer.MAX_VALUE;
525 for( int i = 0; i < linked_nodes.length; ++i ) {
526 final int id_i = linked_nodes[ i ].getId();
527 if ( id_i > max_i_id ) {
529 max_i_id = linked_nodes[ max_i ].getId();
531 if ( id_i < min_i_id ) {
533 min_i_id = linked_nodes[ min_i ].getId();
536 return new int[] { min_i, max_i };
539 * Updates the mapping function M after the root of the gene tree has been
540 * moved by one branch. It calculates M for the root of the gene tree and
541 * one of its two children.
543 * To be used ONLY by method "SDIunrooted.fastInfer(Phylogeny,Phylogeny)".
547 * @param prev_root_was_dup
548 * true if the previous root was a duplication, false otherwise
549 * @param prev_root_c1
550 * child 1 of the previous root
551 * @param prev_root_c2
552 * child 2 of the previous root
553 * @return number of duplications which have been assigned in gene tree
555 // int updateM( final boolean prev_root_was_dup,
556 // final PhylogenyNode prev_root_c1, final PhylogenyNode prev_root_c2 ) {
557 // final PhylogenyNode root = getGeneTree().getRoot();
558 // if ( ( root.getChildNode1() == prev_root_c1 )
559 // || ( root.getChildNode2() == prev_root_c1 ) ) {
560 // calculateMforNode( prev_root_c1 );
563 // calculateMforNode( prev_root_c2 );
565 // Event event = null;
566 // if ( prev_root_was_dup ) {
567 // event = Event.createSingleDuplicationEvent();
570 // event = Event.createSingleSpeciationEvent();
572 // root.getPhylogenyNodeData().setEvent( event );
573 // calculateMforNode( root );
574 // return getDuplications();
575 // } // updateM( boolean, PhylogenyNode, PhylogenyNode )
576 // Helper method for updateM( boolean, PhylogenyNode, PhylogenyNode )
577 // Calculates M for PhylogenyNode n, given that M for the two children
578 // of n has been calculated.
579 // (Last modified: 10/02/01)
580 // private void calculateMforNode( final PhylogenyNode n ) {
581 // if ( !n.isExternal() ) {
582 // boolean was_duplication = n.isDuplication();
583 // PhylogenyNode a = n.getChildNode1().getLink(), b = n
584 // .getChildNode2().getLink();
585 // while ( a != b ) {
586 // if ( a.getID() > b.getID() ) {
587 // a = a.getParent();
590 // b = b.getParent();
594 // Event event = null;
595 // if ( ( a == n.getChildNode1().getLink() )
596 // || ( a == n.getChildNode2().getLink() ) ) {
597 // event = Event.createSingleDuplicationEvent();
598 // if ( !was_duplication ) {
599 // increaseDuplications();
603 // event = Event.createSingleSpeciationEvent();
604 // if ( was_duplication ) {
605 // decreaseDuplications();
608 // n.getPhylogenyNodeData().setEvent( event );
610 // } // calculateMforNode( PhylogenyNode )