inprogress

[jalview.git] / forester / java / src / org / forester / sdi / GSDI.java

diff --git a/forester/java/src/org/forester/sdi/GSDI.java b/forester/java/src/org/forester/sdi/GSDI.java
index fd64970..63c6215 100644 (file)
--- a/forester/java/src/org/forester/sdi/GSDI.java
+++ b/forester/java/src/org/forester/sdi/GSDI.java
@@ -41,7 +41,7 @@ import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
 import org.forester.sdi.SDIutil.TaxonomyComparisonBase;
 import org.forester.util.ForesterUtil;
 
-public final class GSDI implements GSDII  {
+public final class GSDI implements GSDII {
 
     private final boolean                _most_parsimonious_duplication_model;
     private final int                    _speciation_or_duplication_events_sum;
@@ -59,9 +59,11 @@ public final class GSDI implements GSDII  {
                  final boolean strip_gene_tree,
                  final boolean strip_species_tree ) throws SDIException {
         _most_parsimonious_duplication_model = most_parsimonious_duplication_model;
+        if ( gene_tree.getRoot().getNumberOfDescendants() == 3 ) {
+            gene_tree.reRoot( gene_tree.getRoot().getChildNode( 2 ) );
+        }
         final NodesLinkingResult nodes_linking_result = linkNodesOfG( gene_tree,
                                                                       species_tree,
-                                                                      null,
                                                                       strip_gene_tree,
                                                                       strip_species_tree );
         _stripped_gene_tree_nodes = nodes_linking_result.getStrippedGeneTreeNodes();
@@ -82,10 +84,12 @@ public final class GSDI implements GSDII  {
         return _duplications_sum;
     }
 
+    @Override
     public Set<PhylogenyNode> getMappedExternalSpeciesTreeNodes() {
         return _mapped_species_tree_nodes;
     }
 
+    @Override
     public final SortedSet<String> getReMappedScientificNamesFromGeneTree() {
         return _scientific_names_mapped_to_reduced_specificity;
     }
@@ -94,18 +98,22 @@ public final class GSDI implements GSDII  {
         return _speciation_or_duplication_events_sum;
     }
 
+    @Override
     public final int getSpeciationsSum() {
         return _speciations_sum;
     }
 
+    @Override
     public List<PhylogenyNode> getStrippedExternalGeneTreeNodes() {
         return _stripped_gene_tree_nodes;
     }
 
+    @Override
     public List<PhylogenyNode> getStrippedSpeciesTreeNodes() {
         return _stripped_species_tree_nodes;
     }
 
+    @Override
     public TaxonomyComparisonBase getTaxCompBase() {
         return _tax_comp_base;
     }
@@ -135,14 +143,20 @@ public final class GSDI implements GSDII  {
      * the species tree must be labeled in preorder.
      * <p>
      * @return 
+     * @throws SDIException 
      * 
      */
     final static GSDIsummaryResult geneTreePostOrderTraversal( final Phylogeny gene_tree,
-                                                               final boolean most_parsimonious_duplication_model ) {
+                                                               final boolean most_parsimonious_duplication_model )
+            throws SDIException {
         final GSDIsummaryResult res = new GSDIsummaryResult();
         for( final PhylogenyNodeIterator it = gene_tree.iteratorPostorder(); it.hasNext(); ) {
             final PhylogenyNode g = it.next();
             if ( g.isInternal() ) {
+                if ( g.getNumberOfDescendants() != 2 ) {
+                    throw new SDIException( "gene tree contains internal node with " + g.getNumberOfDescendants()
+                            + " descendents" );
+                }
                 PhylogenyNode s1 = g.getChildNode1().getLink();
                 PhylogenyNode s2 = g.getChildNode2().getLink();
                 while ( s1 != s2 ) {
@@ -160,6 +174,17 @@ public final class GSDI implements GSDII  {
         return res;
     }
 
+    final static NodesLinkingResult linkNodesOfG( final Phylogeny gene_tree,
+                                                  final Phylogeny species_tree,
+                                                  final boolean strip_gene_tree,
+                                                  final boolean strip_species_tree ) throws SDIException {
+        final TaxonomyComparisonBase tax_comp_base = SDIutil.determineTaxonomyComparisonBase( gene_tree );
+        if ( tax_comp_base == null ) {
+            throw new RuntimeException( "failed to establish taxonomy linking base (taxonomy linking base is null)" );
+        }
+        return linkNodesOfG( gene_tree, species_tree, tax_comp_base, strip_gene_tree, strip_species_tree );
+    }
+
     /**
      * This allows for linking of internal nodes of the species tree (as opposed
      * to just external nodes, as in the method it overrides.
@@ -172,15 +197,13 @@ public final class GSDI implements GSDII  {
                                                   final TaxonomyComparisonBase tax_comp_base,
                                                   final boolean strip_gene_tree,
                                                   final boolean strip_species_tree ) throws SDIException {
+        if ( tax_comp_base == null ) {
+            throw new IllegalArgumentException( "taxonomy linking base is null" );
+        }
         final Map<String, PhylogenyNode> species_to_node_map = new HashMap<String, PhylogenyNode>();
         final List<PhylogenyNode> species_tree_ext_nodes = new ArrayList<PhylogenyNode>();
         final NodesLinkingResult res = new NodesLinkingResult();
-        if ( tax_comp_base == null ) {
-            res.setTaxCompBase( SDIutil.determineTaxonomyComparisonBase( gene_tree ) );
-        }
-        else {
-            res.setTaxCompBase( tax_comp_base );
-        }
+        res.setTaxCompBase( tax_comp_base );
         // Stringyfied taxonomy is the key, node is the value.
         for( final PhylogenyNodeIterator iter = species_tree.iteratorExternalForward(); iter.hasNext(); ) {
             final PhylogenyNode s = iter.next();
@@ -189,7 +212,8 @@ public final class GSDI implements GSDII  {
                 final String tax_str = SDIutil.taxonomyToString( s, res.getTaxCompBase() );
                 if ( !ForesterUtil.isEmpty( tax_str ) ) {
                     if ( species_to_node_map.containsKey( tax_str ) ) {
-                        throw new SDIException( "taxonomy \"" + s + "\" is not unique in species tree" );
+                        throw new SDIException( "taxonomy \"" + tax_str + "\" is not unique in species tree (using "
+                                + res.getTaxCompBase() + " for linking to gene tree)" );
                     }
                     species_to_node_map.put( tax_str, s );
                 }
@@ -243,7 +267,8 @@ public final class GSDI implements GSDII  {
         if ( strip_gene_tree ) {
             stripTree( gene_tree, res.getStrippedGeneTreeNodes() );
             if ( gene_tree.isEmpty() || ( gene_tree.getNumberOfExternalNodes() < 2 ) ) {
-                throw new SDIException( "species could not be mapped between gene tree and species tree" );
+                throw new SDIException( "species could not be mapped between gene tree and species tree (based on "
+                        + res.getTaxCompBase() + ")" );
             }
         }
         if ( strip_species_tree ) {