phylotastic hackathon at NESCENT 120607

diff --git a/forester/java/src/org/forester/application/gene_tree_preprocess.java b/forester/java/src/org/forester/application/gene_tree_preprocess.java
index 41f6a47..840c335 100644 (file)
--- a/forester/java/src/org/forester/application/gene_tree_preprocess.java
+++ b/forester/java/src/org/forester/application/gene_tree_preprocess.java
@@ -52,7 +52,7 @@ public class gene_tree_preprocess {
     final static private String PRG_VERSION   = "1.01";
     final static private String PRG_DATE      = "2012.06.07";
     final static private String E_MAIL        = "phylosoft@gmail.com";
-    final static private String WWW           = "www.phylosoft.org/forester/";
+    final static private String WWW           = "www.phylosoft.org/forester";
 
     public static void main( final String[] args ) {
         try {
@@ -82,9 +82,8 @@ public class gene_tree_preprocess {
                 ForesterUtil.fatalError( PRG_NAME, "phylogeny has " + phy.getNumberOfExternalNodes()
                         + " external node(s), aborting" );
             }
-            final SortedSet<String> not_found = SequenceDataRetriver.obtainSeqInformation( phy, true );
+            final SortedSet<String> not_found = SequenceDataRetriver.obtainSeqInformation( phy, true , false);
             for( final String remove_me : not_found ) {
-                //PhylogenyMethods.removeNode( phy.getNode( remove_me ), phy );
                 phy.deleteSubtree( phy.getNode( remove_me ), true );
             }
             if ( phy.getNumberOfExternalNodes() < 2 ) {

diff --git a/forester/java/src/org/forester/application/gsdi.java b/forester/java/src/org/forester/application/gsdi.java
index 3763e1b..708b734 100644 (file)
--- a/forester/java/src/org/forester/application/gsdi.java
+++ b/forester/java/src/org/forester/application/gsdi.java
@@ -31,9 +31,13 @@ import java.util.ArrayList;
 import java.util.Date;
 import java.util.List;
 
+import org.forester.archaeopteryx.Archaeopteryx;
+import org.forester.io.parsers.PhylogenyParser;
 import org.forester.io.parsers.phyloxml.PhyloXmlParser;
+import org.forester.io.parsers.util.ParserUtils;
 import org.forester.io.writers.PhylogenyWriter;
 import org.forester.phylogeny.Phylogeny;
+import org.forester.phylogeny.PhylogenyMethods;
 import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
 import org.forester.phylogeny.factories.PhylogenyFactory;
 import org.forester.sdi.GSDI;
@@ -44,18 +48,19 @@ import org.forester.util.ForesterUtil;
 
 public final class gsdi {
 
-    final static private String STRIP_OPTION             = "s";
-    final static private String GSDI_OPTION              = "g";
-    final static private String MOST_PARSIMONIOUS_OPTION = "m";
-    final static private String HELP_OPTION_1            = "help";
-    final static private String HELP_OPTION_2            = "h";
-    final static private String DEFAULT_OUTFILE          = "sdi_out.xml";
-    final static private String PRG_NAME                 = "gsdi";
-    final static private String PRG_VERSION              = "0.4";
-    final static private String PRG_DATE                 = "120607";
-    final static private String PRG_DESC                 = "general speciation duplication inference";
-    final static private String E_MAIL                   = "phylosoft@gmail.com";
-    final static private String WWW                      = "www.phylosoft.org/forester/";
+    final static private String STRIP_OPTION                 = "s";
+    final static private String SDI_OPTION                   = "b";
+    final static private String MOST_PARSIMONIOUS_OPTION     = "m";
+    final static private String GUESS_FORMAT_OF_SPECIES_TREE = "q";
+    final static private String HELP_OPTION_1                = "help";
+    final static private String HELP_OPTION_2                = "h";
+    final static private String DEFAULT_OUTFILE              = "gsdi_out.phylo.xml";
+    final static private String PRG_NAME                     = "gsdi";
+    final static private String PRG_VERSION                  = "0.4";
+    final static private String PRG_DATE                     = "120607";
+    final static private String PRG_DESC                     = "general speciation duplication inference";
+    final static private String E_MAIL                       = "phylosoft@gmail.com";
+    final static private String WWW                          = "www.phylosoft.org/forester";
 
     public static void main( final String args[] ) {
         ForesterUtil.printProgramInformation( PRG_NAME,
@@ -86,20 +91,22 @@ public final class gsdi {
         }
         final List<String> allowed_options = new ArrayList<String>();
         allowed_options.add( gsdi.STRIP_OPTION );
-        allowed_options.add( gsdi.GSDI_OPTION );
+        allowed_options.add( gsdi.SDI_OPTION );
+        allowed_options.add( gsdi.GUESS_FORMAT_OF_SPECIES_TREE );
         allowed_options.add( gsdi.MOST_PARSIMONIOUS_OPTION );
         final String dissallowed_options = cla.validateAllowedOptionsAsString( allowed_options );
         if ( dissallowed_options.length() > 0 ) {
             ForesterUtil.fatalError( gsdi.PRG_NAME, "unknown option(s): " + dissallowed_options );
         }
-        boolean use_sdise = true;
+        boolean use_sdise = false;
         boolean strip = false;
         boolean most_parsimonous_duplication_model = false;
+        boolean species_tree_in_phyloxml = true;
         if ( cla.isOptionSet( gsdi.STRIP_OPTION ) ) {
             strip = true;
         }
-        if ( cla.isOptionSet( gsdi.GSDI_OPTION ) ) {
-            use_sdise = false;
+        if ( cla.isOptionSet( gsdi.SDI_OPTION ) ) {
+            use_sdise = true;
         }
         if ( cla.isOptionSet( gsdi.MOST_PARSIMONIOUS_OPTION ) ) {
             if ( use_sdise ) {
@@ -107,6 +114,9 @@ public final class gsdi {
             }
             most_parsimonous_duplication_model = true;
         }
+        if ( cla.isOptionSet( gsdi.GUESS_FORMAT_OF_SPECIES_TREE ) ) {
+            species_tree_in_phyloxml = false;
+        }
         Phylogeny species_tree = null;
         Phylogeny gene_tree = null;
         File gene_tree_file = null;
@@ -136,12 +146,22 @@ public final class gsdi {
         }
         try {
             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
-            species_tree = factory.create( species_tree_file, new PhyloXmlParser() )[ 0 ];
+            if ( species_tree_in_phyloxml ) {
+                species_tree = factory.create( species_tree_file, new PhyloXmlParser() )[ 0 ];
+            }
+            else {
+                final PhylogenyParser p = ParserUtils.createParserDependingOnFileType( species_tree_file, true );
+                species_tree = factory.create( species_tree_file, p )[ 0 ];
+                PhylogenyMethods.transferNodeNameToField( species_tree,
+                                                          PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME );
+              
+            }
         }
         catch ( final IOException e ) {
             ForesterUtil.fatalError( gsdi.PRG_NAME,
                                      "Failed to read species tree from [" + gene_tree_file + "]: " + e.getMessage() );
         }
+        
         try {
             final PhylogenyFactory factory = ParserBasedPhylogenyFactory.getInstance();
             gene_tree = factory.create( gene_tree_file, new PhyloXmlParser() )[ 0 ];
@@ -153,12 +173,12 @@ public final class gsdi {
         gene_tree.setRooted( true );
         species_tree.setRooted( true );
         if ( !gene_tree.isCompletelyBinary() ) {
-            ForesterUtil.fatalError( gsdi.PRG_NAME, "gene tree (\"" + gene_tree_file + "\") is not completely binary." );
+            ForesterUtil.fatalError( gsdi.PRG_NAME, "gene tree (\"" + gene_tree_file + "\") is not completely binary" );
         }
         if ( use_sdise ) {
             if ( !species_tree.isCompletelyBinary() ) {
                 ForesterUtil.fatalError( gsdi.PRG_NAME, "species tree (\"" + species_tree_file
-                        + "\") is not completely binary." );
+                        + "\") is not completely binary" );
             }
         }
         // For timing.
@@ -180,12 +200,12 @@ public final class gsdi {
         try {
             if ( use_sdise ) {
                 System.out.println();
-                System.out.println( "Using SDIse algorithm." );
+                System.out.println( "Using SDIse algorithm" );
                 sdi = new SDIse( gene_tree, species_tree );
             }
             else {
                 System.out.println();
-                System.out.println( "Using GSDI algorithm." );
+                System.out.println( "Using GSDI algorithm" );
                 System.out.println();
                 System.out.println( "Use most parsimonous duplication model: " + most_parsimonous_duplication_model );
                 sdi = new GSDI( gene_tree, species_tree, most_parsimonous_duplication_model );
@@ -201,7 +221,7 @@ public final class gsdi {
             writer.toPhyloXML( out_file, gene_tree, 0 );
         }
         catch ( final IOException e ) {
-            ForesterUtil.fatalError( PRG_NAME, "Failed to write to \"" + out_file + "\" [" + e.getMessage() + "]" );
+            ForesterUtil.fatalError( PRG_NAME, "Failed to write to [" + out_file + "]: " + e.getMessage() );
         }
         System.out.println();
         System.out.println( "Successfully wrote resulting gene tree to: " + out_file );
@@ -219,28 +239,30 @@ public final class gsdi {
             System.out.println( "Number speciations              : " + ( ( GSDI ) sdi ).getSpeciationsSum() );
         }
         System.out.println();
-    } 
+    }
 
     private static void print_help() {
-        System.out.println( "Usage: \"" + gsdi.PRG_NAME
-                + " [-options] <gene tree in phyloXML format> <species tree in phyloXML format> [outfile]\"" );
+        System.out.println( "Usage: " + gsdi.PRG_NAME
+                + " [-options] <gene tree in phyloXML format> <species tree in phyloXML format> [outfile]" );
         System.out.println();
         System.out.println( "Options:" );
         System.out.println( " -" + gsdi.STRIP_OPTION + ": to strip the species tree prior to duplication inference" );
-        System.out.println( " -" + gsdi.GSDI_OPTION
-                + ": to use GSDI algorithm instead of SDIse algorithm" );
-        System.out
-                .println( " -" + gsdi.MOST_PARSIMONIOUS_OPTION + ": use most parimonious duplication model for GSDI: " );
+        System.out.println( " -" + gsdi.SDI_OPTION + ": to use SDI algorithm instead of GSDI algorithm" );
+        System.out.println( " -" + gsdi.MOST_PARSIMONIOUS_OPTION
+                + ": use most parimonious duplication model for GSDI: " );
         System.out.println( "     assign nodes as speciations which would otherwise be assiged" );
         System.out.println( "     as unknown because of polytomies in the species tree" );
+        System.out.println( " -" + gsdi.GUESS_FORMAT_OF_SPECIES_TREE + ": to allow species tree in other formats than" );
+        System.out.println( "     phyloXML (Newick, NHX, Nexus)" );
         System.out.println();
         System.out.println( "Species tree:" );
-        System.out.println( " In phyloXML format, with taxonomy data in appropriate fields." );
+        System.out.println( " In phyloXML format (unless option " + gsdi.GUESS_FORMAT_OF_SPECIES_TREE
+                + " is used), with taxonomy data in appropriate fields" );
         System.out.println();
         System.out.println( "Gene tree:" );
-        System.out.println( " In phyloXM format, with taxonomy and sequence data in appropriate fields." );
+        System.out.println( " In phyloXM format, with taxonomy and sequence data in appropriate fields" );
         System.out.println();
-        System.out.println( "Note. GSDI algorithm is under development." );
+        System.out.println( "Note -- GSDI algorithm is under development" );
         System.out.println();
     }
 }

diff --git a/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java b/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
index 1ac188a..7c6291d 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
+++ b/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
@@ -68,7 +68,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
         start( _mf, "sequence data" );
         SortedSet<String> not_found = null;
         try {
-            not_found = obtainSeqInformation( _phy, false );
+            not_found = obtainSeqInformation( _phy, false, true );
         }
         catch ( final UnknownHostException e ) {
             final String what = "_"; //TODO FIXME 
@@ -142,7 +142,8 @@ public final class SequenceDataRetriver extends RunnableProcess {
         }
     }
 
-    public static SortedSet<String> obtainSeqInformation( final Phylogeny phy, final boolean ext_nodes_only )
+    public static SortedSet<String> obtainSeqInformation( final Phylogeny phy, final boolean ext_nodes_only,
+                                                          final boolean allow_to_set_taxonomic_data )
             throws IOException {
         final SortedSet<String> not_found = new TreeSet<String>();
         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
@@ -237,7 +238,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
                 if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
                     tax.setScientificName( db_entry.getTaxonomyScientificName() );
                 }
-                if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
+                if ( allow_to_set_taxonomic_data && !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
                     tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
                 }
                 node.getNodeData().setTaxonomy( tax );

diff --git a/forester/java/src/org/forester/phylogeny/data/Taxonomy.java b/forester/java/src/org/forester/phylogeny/data/Taxonomy.java
index d403a8e..1375652 100644 (file)
--- a/forester/java/src/org/forester/phylogeny/data/Taxonomy.java
+++ b/forester/java/src/org/forester/phylogeny/data/Taxonomy.java
@@ -197,7 +197,7 @@ public class Taxonomy implements PhylogenyData, MultipleUris, Comparable<Taxonom
 
     @Override
     public int hashCode() {
-        if ( getIdentifier() != null ) {
+        if ( getIdentifier() != null && !ForesterUtil.isEmpty( getIdentifier().getValue() ) ) {
             return getIdentifier().hashCode();
         }
         else if ( !ForesterUtil.isEmpty( getTaxonomyCode() ) ) {
@@ -251,7 +251,11 @@ public class Taxonomy implements PhylogenyData, MultipleUris, Comparable<Taxonom
             return true;
         }
         final Taxonomy tax = ( Taxonomy ) data;
-        if ( ( getIdentifier() != null ) && ( tax.getIdentifier() != null ) ) {
+        if ( ( getIdentifier() != null ) && ( tax.getIdentifier() != null )
+          && !ForesterUtil.isEmpty( getIdentifier().getValue() )  
+          && !ForesterUtil.isEmpty( tax.getIdentifier().getValue() )
+        
+        ) {
             return getIdentifier().isEqual( tax.getIdentifier() );
         }
         else if ( !ForesterUtil.isEmpty( getTaxonomyCode() ) && !ForesterUtil.isEmpty( tax.getTaxonomyCode() ) ) {

diff --git a/forester/java/src/org/forester/sdi/GSDI.java b/forester/java/src/org/forester/sdi/GSDI.java
index 9b18e81..58f79fa 100644 (file)
--- a/forester/java/src/org/forester/sdi/GSDI.java
+++ b/forester/java/src/org/forester/sdi/GSDI.java
@@ -56,7 +56,7 @@ import org.forester.util.ForesterUtil;
  * 
  * @author Christian M. Zmasek
  */
-public class GSDI extends SDI {
+public final class GSDI extends SDI {
 
     private final HashMap<PhylogenyNode, Integer> _transversal_counts;
     private final boolean                         _most_parsimonious_duplication_model;
@@ -103,26 +103,26 @@ public class GSDI extends SDI {
         geneTreePostOrderTraversal( getGeneTree().getRoot() );
     }
 
-    private Event createDuplicationEvent() {
+    private final Event createDuplicationEvent() {
         final Event event = Event.createSingleDuplicationEvent();
         ++_duplications_sum;
         return event;
     }
 
-    private Event createSingleSpeciationOrDuplicationEvent() {
+    private final Event createSingleSpeciationOrDuplicationEvent() {
         final Event event = Event.createSingleSpeciationOrDuplicationEvent();
         ++_speciation_or_duplication_events_sum;
         return event;
     }
 
-    private Event createSpeciationEvent() {
+    private final Event createSpeciationEvent() {
         final Event event = Event.createSingleSpeciationEvent();
         ++_speciations_sum;
         return event;
     }
 
     // s is the node on the species tree g maps to.
-    private void determineEvent( final PhylogenyNode s, final PhylogenyNode g ) {
+    private final void determineEvent( final PhylogenyNode s, final PhylogenyNode g ) {
         Event event = null;
         // Determine how many children map to same node as parent.
         int sum_g_childs_mapping_to_s = 0;
@@ -190,7 +190,7 @@ public class GSDI extends SDI {
      * @param g
      *            starting node of a gene tree - normally the root
      */
-    void geneTreePostOrderTraversal( final PhylogenyNode g ) {
+    final void geneTreePostOrderTraversal( final PhylogenyNode g ) {
         if ( !g.isExternal() ) {
             for( final PhylogenyNodeIterator iter = g.iterateChildNodesForward(); iter.hasNext(); ) {
                 geneTreePostOrderTraversal( iter.next() );
@@ -218,22 +218,22 @@ public class GSDI extends SDI {
         }
     }
 
-    public int getSpeciationOrDuplicationEventsSum() {
+    public final int getSpeciationOrDuplicationEventsSum() {
         return _speciation_or_duplication_events_sum;
     }
 
-    public int getSpeciationsSum() {
+    public final int getSpeciationsSum() {
         return _speciations_sum;
     }
 
-    private int getTraversalCount( final PhylogenyNode node ) {
+    private final int getTraversalCount( final PhylogenyNode node ) {
         if ( _transversal_counts.containsKey( node ) ) {
             return _transversal_counts.get( node );
         }
         return 0;
     }
 
-    private void increaseTraversalCount( final PhylogenyNode node ) {
+    private final void increaseTraversalCount( final PhylogenyNode node ) {
         if ( _transversal_counts.containsKey( node ) ) {
             _transversal_counts.put( node, _transversal_counts.get( node ) + 1 );
         }
@@ -250,7 +250,7 @@ public class GSDI extends SDI {
      * 
      */
     @Override
-    void linkNodesOfG() {
+    final  void linkNodesOfG() {
         final HashMap<Taxonomy, PhylogenyNode> speciestree_ext_nodes = new HashMap<Taxonomy, PhylogenyNode>();
         for( final PhylogenyNodeIterator iter = _species_tree.iteratorLevelOrder(); iter.hasNext(); ) {
             final PhylogenyNode n = iter.next();
@@ -272,14 +272,16 @@ public class GSDI extends SDI {
             final PhylogenyNode s = speciestree_ext_nodes.get( g.getNodeData().getTaxonomy() );
             if ( s == null ) {
                 throw new IllegalArgumentException( "species " + g.getNodeData().getTaxonomy()
-                        + " not present in species tree." );
+                        + " not present in species tree" );
             }
+          
+          
             g.setLink( s );
         }
     }
 
     @Override
-    public String toString() {
+    public final String toString() {
         final StringBuffer sb = new StringBuffer();
         sb.append( "Most parsimonious duplication model: " + _most_parsimonious_duplication_model );
         sb.append( ForesterUtil.getLineSeparator() );
@@ -295,7 +297,7 @@ public class GSDI extends SDI {
         return sb.toString();
     }
 
-    static int[] obtainMinMaxIdIndices( final PhylogenyNode[] linked_nodes ) {
+    static final int[] obtainMinMaxIdIndices( final PhylogenyNode[] linked_nodes ) {
         int max_i = 0;
         int min_i = 0;
         int max_i_id = -Integer.MAX_VALUE;
@@ -386,4 +388,4 @@ public class GSDI extends SDI {
     // n.getPhylogenyNodeData().setEvent( event );
     // }
     // } // calculateMforNode( PhylogenyNode )
-} // End of class GSDI.
+}