in progress

diff --git a/forester/java/src/org/forester/analysis/AncestralTaxonomyInference.java b/forester/java/src/org/forester/analysis/AncestralTaxonomyInference.java
index 4f114a2..7ddfe99 100644 (file)
--- a/forester/java/src/org/forester/analysis/AncestralTaxonomyInference.java
+++ b/forester/java/src/org/forester/analysis/AncestralTaxonomyInference.java
@@ -275,10 +275,15 @@ public final class AncestralTaxonomyInference {
                 .equalsIgnoreCase( "uniprotkb" ) ) ) );
     }
 
-    synchronized public static SortedSet<String> obtainDetailedTaxonomicInformation( final Phylogeny phy )
+    synchronized public static SortedSet<String> obtainDetailedTaxonomicInformation( final Phylogeny phy,
+                                                                                     final boolean delete )
             throws IOException {
         clearCachesIfTooLarge();
         final SortedSet<String> not_found = new TreeSet<String>();
+        List<PhylogenyNode> not_found_external_nodes = null;
+        if ( delete ) {
+            not_found_external_nodes = new ArrayList<PhylogenyNode>();
+        }
         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
             final PhylogenyNode node = iter.next();
             final QUERY_TYPE qt = null;
@@ -293,6 +298,9 @@ public final class AncestralTaxonomyInference {
                 else {
                     not_found.add( node.toString() );
                 }
+                if ( delete ) {
+                    not_found_external_nodes.add( node );
+                }
             }
             UniProtTaxonomy up_tax = null;
             if ( ( tax != null )
@@ -305,9 +313,18 @@ public final class AncestralTaxonomyInference {
                 }
                 else {
                     not_found.add( tax.toString() );
+                    if ( delete && node.isExternal() ) {
+                        not_found_external_nodes.add( node );
+                    }
                 }
             }
         }
+        if ( delete ) {
+            for( PhylogenyNode node : not_found_external_nodes ) {
+                phy.deleteSubtree( node, false );
+            }
+            phy.recalculateNumberOfExternalDescendants( true );
+        }
         return not_found;
     }
 

diff --git a/forester/java/src/org/forester/application/get_subtree_specific_chars.java b/forester/java/src/org/forester/application/get_subtree_specific_chars.java
index 830b043..2f0f143 100644 (file)
--- a/forester/java/src/org/forester/application/get_subtree_specific_chars.java
+++ b/forester/java/src/org/forester/application/get_subtree_specific_chars.java
@@ -61,8 +61,12 @@ public class get_subtree_specific_chars {
         final SortedSet<String> all_chars = getAllExternalPresentAndGainedCharacters( phy.getRoot() );
         System.out.println( "Sum of all external characters:\t" + all_chars.size() );
         System.out.println();
+        final boolean SIMPLE = false;
         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
             final PhylogenyNode node = iter.next();
+            if ( !SIMPLE && node.isExternal() ) {
+                continue;
+            }
             if ( !node.isRoot() ) {
                 System.out.println();
                 if ( node.getNodeData().isHasTaxonomy()
@@ -84,7 +88,24 @@ public class get_subtree_specific_chars {
                 final SortedSet<String> unique_chars = new TreeSet<String>();
                 for( final String node_char : node_chars ) {
                     if ( !not_node_chars.contains( node_char ) ) {
-                        unique_chars.add( node_char );
+                        if ( SIMPLE ) {
+                            unique_chars.add( node_char );
+                        }
+                        else {
+                            boolean found = true;
+                            for( final int external_id : external_ids ) {
+                                if ( !phy.getNode( external_id ).getNodeData().getBinaryCharacters()
+                                        .getGainedCharacters().contains( node_char )
+                                        && !phy.getNode( external_id ).getNodeData().getBinaryCharacters()
+                                                .getPresentCharacters().contains( node_char ) ) {
+                                    found = false;
+                                    break;
+                                }
+                            }
+                            if ( found ) {
+                                unique_chars.add( node_char );
+                            }
+                        }
                     }
                 }
                 System.out.println( "\tSUM:\t" + unique_chars.size() );

diff --git a/forester/java/src/org/forester/archaeopteryx/MainFrame.java b/forester/java/src/org/forester/archaeopteryx/MainFrame.java
index 80555c0..5e78865 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/MainFrame.java
+++ b/forester/java/src/org/forester/archaeopteryx/MainFrame.java
@@ -120,6 +120,7 @@ public abstract class MainFrame extends JFrame implements ActionListener {
     JMenuItem                 _collapse_species_specific_subtrees;
     JMenuItem                 _collapse_below_threshold;                                                                                                                                                                                //TODO implememt me
     JMenuItem                 _obtain_detailed_taxonomic_information_jmi;
+    JMenuItem                 _obtain_detailed_taxonomic_information_deleting_jmi;
     JMenuItem                 _obtain_uniprot_seq_information_jmi;
     JMenuItem                 _move_node_names_to_tax_sn_jmi;
     JMenuItem                 _move_node_names_to_seq_names_jmi;

diff --git a/forester/java/src/org/forester/archaeopteryx/MainFrameApplication.java b/forester/java/src/org/forester/archaeopteryx/MainFrameApplication.java
index 461f9f3..80c3b5b 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/MainFrameApplication.java
+++ b/forester/java/src/org/forester/archaeopteryx/MainFrameApplication.java
@@ -496,6 +496,12 @@ public final class MainFrameApplication extends MainFrame {
                 }
                 obtainDetailedTaxonomicInformation();
             }
+            else if ( o == _obtain_detailed_taxonomic_information_deleting_jmi ) {
+                if ( isSubtreeDisplayed() ) {
+                    return;
+                }
+                obtainDetailedTaxonomicInformationDelete();
+            }
             else if ( o == _obtain_uniprot_seq_information_jmi ) {
                 obtainUniProtSequenceInformation();
             }
@@ -870,6 +876,11 @@ public final class MainFrameApplication extends MainFrame {
         _obtain_detailed_taxonomic_information_jmi
                 .setToolTipText( "To add additional taxonomic information (from UniProt Taxonomy)" );
         _tools_menu
+                .add( _obtain_detailed_taxonomic_information_deleting_jmi = new JMenuItem( "Obtain Detailed Taxonomic Information (Delete Nodes)" ) );
+        customizeJMenuItem( _obtain_detailed_taxonomic_information_deleting_jmi );
+        _obtain_detailed_taxonomic_information_deleting_jmi
+                .setToolTipText( "To add additional taxonomic information, deletes nodes for which taxonomy cannot found (from UniProt Taxonomy)" );
+        _tools_menu
                 .add( _obtain_uniprot_seq_information_jmi = new JMenuItem( "Obtain Sequence Information (from UniProt)" ) );
         customizeJMenuItem( _obtain_uniprot_seq_information_jmi );
         _obtain_uniprot_seq_information_jmi.setToolTipText( "To add additional sequence information (from UniProt)" );
@@ -1427,6 +1438,19 @@ public final class MainFrameApplication extends MainFrame {
         }
     }
 
+    private void obtainDetailedTaxonomicInformationDelete() {
+        if ( getCurrentTreePanel() != null ) {
+            final Phylogeny phy = getCurrentTreePanel().getPhylogeny();
+            if ( ( phy != null ) && !phy.isEmpty() ) {
+                final TaxonomyDataObtainer t = new TaxonomyDataObtainer( this,
+                                                                         _mainpanel.getCurrentTreePanel(),
+                                                                         phy.copy(),
+                                                                         true );
+                new Thread( t ).start();
+            }
+        }
+    }
+
     private void obtainUniProtSequenceInformation() {
         if ( getCurrentTreePanel() != null ) {
             final Phylogeny phy = getCurrentTreePanel().getPhylogeny();

diff --git a/forester/java/src/org/forester/archaeopteryx/tools/TaxonomyDataObtainer.java b/forester/java/src/org/forester/archaeopteryx/tools/TaxonomyDataObtainer.java
index 17d871b..952d3fa 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/tools/TaxonomyDataObtainer.java
+++ b/forester/java/src/org/forester/archaeopteryx/tools/TaxonomyDataObtainer.java
@@ -43,11 +43,23 @@ public class TaxonomyDataObtainer implements Runnable {
     private final Phylogeny            _phy;
     private final MainFrameApplication _mf;
     private final TreePanel            _treepanel;
+    private final boolean              _delete;
+
+    public TaxonomyDataObtainer( final MainFrameApplication mf,
+                                 final TreePanel treepanel,
+                                 final Phylogeny phy,
+                                 boolean delete ) {
+        _phy = phy;
+        _mf = mf;
+        _treepanel = treepanel;
+        _delete = delete;
+    }
 
     public TaxonomyDataObtainer( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
         _phy = phy;
         _mf = mf;
         _treepanel = treepanel;
+        _delete = false;
     }
 
     private String getBaseUrl() {
@@ -58,7 +70,7 @@ public class TaxonomyDataObtainer implements Runnable {
         _mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
         SortedSet<String> not_found = null;
         try {
-            not_found = AncestralTaxonomyInference.obtainDetailedTaxonomicInformation( _phy );
+            not_found = AncestralTaxonomyInference.obtainDetailedTaxonomicInformation( _phy, _delete );
         }
         catch ( final UnknownHostException e ) {
             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
@@ -93,10 +105,21 @@ public class TaxonomyDataObtainer implements Runnable {
             final StringBuffer sb = new StringBuffer();
             sb.append( "Not all taxonomies could be resolved.\n" );
             if ( not_found.size() == 1 ) {
-                sb.append( "The following taxonomy was not found:\n" );
+                if ( _delete ) {
+                    sb.append( "The following taxonomy was not found and deleted (if external):\n" );
+                }
+                else {
+                    sb.append( "The following taxonomy was not found:\n" );
+                }
             }
             else {
-                sb.append( "The following taxonomies were not found (total: " + not_found.size() + "):\n" );
+                if ( _delete ) {
+                    sb.append( "The following taxonomies were not found and deleted (if external) (total: "
+                            + not_found.size() + "):\n" );
+                }
+                else {
+                    sb.append( "The following taxonomies were not found (total: " + not_found.size() + "):\n" );
+                }
             }
             int i = 0;
             for( final String string : not_found ) {