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 46e5bee..e9804a3 100644 (file)
--- a/forester/java/src/org/forester/application/gene_tree_preprocess.java
+++ b/forester/java/src/org/forester/application/gene_tree_preprocess.java
@@ -25,7 +25,9 @@
 
 package org.forester.application;
 
+import java.io.BufferedWriter;
 import java.io.File;
+import java.io.FileWriter;
 import java.io.IOException;
 import java.util.SortedSet;
 
@@ -43,10 +45,6 @@ public class gene_tree_preprocess {
 
     final static private String HELP_OPTION_1 = "help";
     final static private String HELP_OPTION_2 = "h";
-    final static private String FROM_OPTION   = "f";
-    final static private String TO_OPTION     = "t";
-    final static private String STEP_OPTION   = "s";
-    final static private String WINDOW_OPTION = "w";
     final static private String PRG_NAME      = "gene_tree_preprocess";
     final static private String PRG_DESC      = "gene tree preprocessing for SDI analysis";
     final static private String PRG_VERSION   = "1.00";
@@ -81,9 +79,9 @@ 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 );
+            final SortedSet<String> not_found = SequenceDataRetriver.obtainSeqInformation( phy, true );
             for( final String remove_me : not_found ) {
-                System.out.println( " not found: " + not_found );
+                System.out.println( " not found: " + remove_me );
                 PhylogenyMethods.removeNode( phy.getNode( remove_me ), phy );
             }
             if ( phy.getNumberOfExternalNodes() < 2 ) {
@@ -98,9 +96,21 @@ public class gene_tree_preprocess {
             catch ( final IOException e ) {
                 ForesterUtil.fatalError( PRG_NAME, "failed to write to [" + outtree + "]: " + e.getLocalizedMessage() );
             }
-            //  ForesterUtil.programMessage( PRG_NAME, "wrote output to: [" + outfile + "]" );
+            ForesterUtil.programMessage( PRG_NAME, "wrote output phylogeny to: " + outtree );
+            try {
+                final BufferedWriter out = new BufferedWriter( new FileWriter( removed_nodes ) );
+                for( final String remove_me : not_found ) {
+                    out.write( remove_me );
+                    out.newLine();
+                }
+                out.close();
+            }
+            catch ( final IOException e ) {
+                ForesterUtil.fatalError( PRG_NAME,
+                                         "failed to write to [" + removed_nodes + "]: " + e.getLocalizedMessage() );
+            }
+            ForesterUtil.programMessage( PRG_NAME, "wrote removed external nodes labels to: " + removed_nodes );
             ForesterUtil.programMessage( PRG_NAME, "OK" );
-            System.out.println();
         }
         catch ( final Exception e ) {
             ForesterUtil.fatalError( PRG_NAME, e.getMessage() );
@@ -122,19 +132,8 @@ public class gene_tree_preprocess {
                                               E_MAIL,
                                               WWW,
                                               ForesterUtil.getForesterLibraryInformation() );
-        System.out.println( "Usage:" );
-        System.out.println();
-        System.out.println( PRG_NAME + " <options> <msa input file>" );
-        System.out.println();
-        System.out.println( " options: " );
-        System.out.println();
-        System.out.println( "   -" + FROM_OPTION + "=<integer>: from (msa column)" );
-        System.out.println( "   -" + TO_OPTION + "=<integer>: to (msa column)" );
-        System.out.println( "    or" );
-        System.out.println( "   -" + WINDOW_OPTION + "=<integer>: window size (msa columns)" );
-        System.out.println( "   -" + STEP_OPTION + "=<integer>: step size (msa columns)" );
-        System.out.println();
-        System.out.println();
+        System.out.print( "Usage: " );
+        System.out.println( PRG_NAME + " <input phylogeny file>" );
         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 b5a8968..ccef986 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
+++ b/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
@@ -47,13 +47,12 @@ import org.forester.ws.seqdb.SequenceDatabaseEntry;
 import org.forester.ws.seqdb.SequenceDbWsTools;
 
 public final class SequenceDataRetriver extends RunnableProcess {
-    
-    public final static int DEFAULT_LINES_TO_RETURN = 50;
 
+    public final static int            DEFAULT_LINES_TO_RETURN = 50;
     private final Phylogeny            _phy;
     private final MainFrameApplication _mf;
     private final TreePanel            _treepanel;
-    private final static boolean       DEBUG = true;
+    private final static boolean       DEBUG                   = false;
 
     private enum Db {
         UNIPROT, EMBL, NCBI, NONE, REFSEQ;
@@ -69,7 +68,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
         start( _mf, "sequence data" );
         SortedSet<String> not_found = null;
         try {
-            not_found = obtainSeqInformation( _phy );
+            not_found = obtainSeqInformation( _phy, false );
         }
         catch ( final UnknownHostException e ) {
             final String what = "_"; //TODO FIXME 
@@ -143,14 +142,15 @@ public final class SequenceDataRetriver extends RunnableProcess {
         }
     }
 
-    public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
+    public static SortedSet<String> obtainSeqInformation( final Phylogeny phy, final boolean ext_nodes_only ) throws IOException {
         final SortedSet<String> not_found = new TreeSet<String>();
         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
             final PhylogenyNode node = iter.next();
-            final Sequence seq = node.getNodeData().isHasSequence() ?  node.getNodeData().getSequence() : new Sequence() ;
-            final Taxonomy tax = node.getNodeData().isHasTaxonomy() ? node.getNodeData().getTaxonomy() : new Taxonomy() ;
-            
-           
+            if ( ext_nodes_only && node.isInternal() ) {
+                continue;
+            }
+            final Sequence seq = node.getNodeData().isHasSequence() ? node.getNodeData().getSequence() : new Sequence();
+            final Taxonomy tax = node.getNodeData().isHasTaxonomy() ? node.getNodeData().getTaxonomy() : new Taxonomy();
             String query = null;
             Identifier id = null;
             Db db = Db.NONE;
@@ -175,16 +175,19 @@ public final class SequenceDataRetriver extends RunnableProcess {
                     db = Db.UNIPROT;
                 }
                 else if ( ( id = SequenceIdParser.parse( node.getName() ) ) != null ) {
-                    
                     if ( id.getProvider().equalsIgnoreCase( Identifier.NCBI ) ) {
                         db = Db.NCBI;
                     }
                     else if ( id.getProvider().equalsIgnoreCase( Identifier.REFSEQ ) ) {
                         db = Db.REFSEQ;
                     }
-                   
                 }
             }
+            
+            if ( db == Db.NONE ) {
+                not_found.add( node.getName() );
+            }
+            
             SequenceDatabaseEntry db_entry = null;
             if ( !ForesterUtil.isEmpty( query ) ) {
                 if ( db == Db.UNIPROT ) {
@@ -197,7 +200,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
                     if ( DEBUG ) {
                         System.out.println( "embl: " + query );
                     }
-                    db_entry = SequenceDbWsTools.obtainEmblEntry(  new Identifier( query ), DEFAULT_LINES_TO_RETURN );
+                    db_entry = SequenceDbWsTools.obtainEmblEntry( new Identifier( query ), DEFAULT_LINES_TO_RETURN );
                     if ( ( db == Db.UNIPROT ) && ( db_entry != null ) ) {
                         db = Db.EMBL;
                     }
@@ -209,6 +212,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
             else if ( ( db == Db.NCBI ) && ( id != null ) ) {
                 db_entry = SequenceDbWsTools.obtainEmblEntry( id, DEFAULT_LINES_TO_RETURN );
             }
+           
             if ( ( db_entry != null ) && !db_entry.isEmpty() ) {
                 if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
                     String type = null;