phylotastic hackathon at NESCENT 120606

diff --git a/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java b/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
index 4db8175..21dcf1f 100644 (file)
--- a/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
+++ b/forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java
@@ -43,6 +43,7 @@ import org.forester.phylogeny.data.Sequence;
 import org.forester.phylogeny.data.Taxonomy;
 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
 import org.forester.util.ForesterUtil;
+import org.forester.util.SequenceIdParser;
 import org.forester.ws.uniprot.DatabaseTools;
 import org.forester.ws.uniprot.SequenceDatabaseEntry;
 import org.forester.ws.uniprot.UniProtWsTools;
@@ -55,7 +56,7 @@ public final class SequenceDataRetriver extends RunnableProcess {
     private final static boolean       DEBUG = false;
 
     private enum Db {
-        UNKNOWN, UNIPROT, EMBL;
+        UNKNOWN, UNIPROT, EMBL, NCBI;
     }
 
     public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
@@ -185,8 +186,8 @@ public final class SequenceDataRetriver extends RunnableProcess {
                 if ( ( query = UniProtWsTools.parseUniProtAccessor( node.getName() ) ) != null ) {
                     db = Db.UNIPROT;
                 }
-                else if ( ( query = DatabaseTools.parseGenbankAccessor( node.getName() ) ) != null ) {
-                    db = Db.EMBL;
+                else if ( ( query = SequenceIdParser.parseGenbankAccessor( node.getName() ) ) != null ) {
+                    db = Db.NCBI;
                 }
             }
             if ( !ForesterUtil.isEmpty( query ) ) {

diff --git a/forester/java/src/org/forester/test/Test.java b/forester/java/src/org/forester/test/Test.java
index b1b35ee..90430aa 100644 (file)
--- a/forester/java/src/org/forester/test/Test.java
+++ b/forester/java/src/org/forester/test/Test.java
@@ -7959,43 +7959,43 @@ public final class Test {
         //Nucleotide: 1 letter + 5 numerals OR 2 letters + 6 numerals
         //Protein:    3 letters + 5 numerals
         //http://www.ncbi.nlm.nih.gov/Sequin/acc.html
-        if ( !DatabaseTools.parseGenbankAccessor( "AY423861" ).equals( "AY423861" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( "AY423861" ).equals( "AY423861" ) ) {
             return false;
         }
-        if ( !DatabaseTools.parseGenbankAccessor( ".AY423861." ).equals( "AY423861" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( ".AY423861." ).equals( "AY423861" ) ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "AAY423861" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "AAY423861" ) != null ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "AY4238612" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "AY4238612" ) != null ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "AAY4238612" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "AAY4238612" ) != null ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "Y423861" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "Y423861" ) != null ) {
             return false;
         }
-        if ( !DatabaseTools.parseGenbankAccessor( "S12345" ).equals( "S12345" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( "S12345" ).equals( "S12345" ) ) {
             return false;
         }
-        if ( !DatabaseTools.parseGenbankAccessor( "|S12345|" ).equals( "S12345" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( "|S12345|" ).equals( "S12345" ) ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "|S123456" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "|S123456" ) != null ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "ABC123456" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "ABC123456" ) != null ) {
             return false;
         }
-        if ( !DatabaseTools.parseGenbankAccessor( "ABC12345" ).equals( "ABC12345" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( "ABC12345" ).equals( "ABC12345" ) ) {
             return false;
         }
-        if ( !DatabaseTools.parseGenbankAccessor( "&ABC12345&" ).equals( "ABC12345" ) ) {
+        if ( !SequenceIdParser.parseGenbankAccessor( "&ABC12345&" ).equals( "ABC12345" ) ) {
             return false;
         }
-        if ( DatabaseTools.parseGenbankAccessor( "ABCD12345" ) != null ) {
+        if ( SequenceIdParser.parseGenbankAccessor( "ABCD12345" ) != null ) {
             return false;
         }
         return true;
@@ -8929,7 +8929,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "ADF31344" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -8942,7 +8942,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "ADF31344" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -8955,7 +8955,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "ADF31344" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -8969,7 +8969,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "AAA96518" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -8982,7 +8982,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "EHB07727" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -8995,7 +8995,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "BAF37827" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );
@@ -9008,7 +9008,7 @@ public final class Test {
                  || ForesterUtil.isEmpty( id.getValue() )
                  || ForesterUtil.isEmpty( id.getProvider() )
                  || !id.getValue().equals( "CAA73223" )
-                 || !id.getProvider().equals( "genbank" ) ) {
+                 || !id.getProvider().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
                     System.out.println( "provider=" + id.getProvider() );

diff --git a/forester/java/src/org/forester/util/SequenceIdParser.java b/forester/java/src/org/forester/util/SequenceIdParser.java
index 7486d8a..8def260 100644 (file)
--- a/forester/java/src/org/forester/util/SequenceIdParser.java
+++ b/forester/java/src/org/forester/util/SequenceIdParser.java
@@ -58,20 +58,75 @@ public final class SequenceIdParser {
                                                                          .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}\\d{6})(?:[^a-zA-Z0-9]|\\Z)" );\r
     private final static Pattern GENBANK_PROTEIN_AC_PATTERN      = Pattern\r
                                                                          .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5})(?:[^a-zA-Z0-9]|\\Z)" );\r
-    private final static boolean DEBUG                           = false;\r
+    private final static boolean DEBUG                           = true;\r
 \r
     \r
-    \r
+    /**\r
+     * Returns null if no match.\r
+     * \r
+     */\r
     public final static Identifier parse( final String s ) {\r
-        String v = DatabaseTools.parseGenbankAccessor( s );\r
+        String v = parseGenbankAccessor( s );\r
+        if ( !ForesterUtil.isEmpty( v ) ) {\r
+            return new Identifier( v, "ncbi" );\r
+        }\r
+        v = parseRefSeqAccessor( s );\r
         if ( !ForesterUtil.isEmpty( v ) ) {\r
-            return new Identifier( v, "genbank" );\r
+            return new Identifier( v, "ncbi" );\r
         }\r
-        \r
         return null;\r
     }\r
     \r
+    /**\r
+     * Returns null if no match.\r
+     * \r
+     * @param query\r
+     * @param db \r
+     * @return\r
+     */\r
+    static public String parseGenbankAccessor( final String query ) {\r
+        Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );\r
+        if ( m.lookingAt() ) {\r
+            return m.group( 1 );\r
+        }\r
+        else {\r
+            m = GENBANK_NUCLEOTIDE_AC_PATTERN_2.matcher( query );\r
+            if ( m.lookingAt() ) {\r
+                return m.group( 1 );\r
+            }\r
+            else {\r
+                m = GENBANK_PROTEIN_AC_PATTERN.matcher( query );\r
+                if ( m.lookingAt() ) {\r
+                    return m.group( 1 );\r
+                }\r
+                else {\r
+                    return null;\r
+                }\r
+            }\r
+        }\r
+    }\r
     \r
+    public final static String parseRefSeqAccessor( final String query ) {\r
+        Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );\r
+        if ( m.lookingAt() ) {\r
+            return m.group( 1 );\r
+        }\r
+        else {\r
+            m = GENBANK_NUCLEOTIDE_AC_PATTERN_2.matcher( query );\r
+            if ( m.lookingAt() ) {\r
+                return m.group( 1 );\r
+            }\r
+            else {\r
+                m = GENBANK_PROTEIN_AC_PATTERN.matcher( query );\r
+                if ( m.lookingAt() ) {\r
+                    return m.group( 1 );\r
+                }\r
+                else {\r
+                    return null;\r
+                }\r
+            }\r
+        }\r
+    }\r
     \r
     \r
     \r

diff --git a/forester/java/src/org/forester/ws/uniprot/DatabaseTools.java b/forester/java/src/org/forester/ws/uniprot/DatabaseTools.java
index 77b317d..d30830c 100644 (file)
--- a/forester/java/src/org/forester/ws/uniprot/DatabaseTools.java
+++ b/forester/java/src/org/forester/ws/uniprot/DatabaseTools.java
@@ -1,52 +1,10 @@
 
 package org.forester.ws.uniprot;
 
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
 
 public class DatabaseTools {
 
-    //The format for GenBank Accession numbers are:
-    //Nucleotide: 1 letter + 5 numerals OR 2 letters + 6 numerals
-    //Protein:    3 letters + 5 numerals
-    //http://www.ncbi.nlm.nih.gov/Sequin/acc.html
-    private final static Pattern GENBANK_NUCLEOTIDE_AC_PATTERN_1 = Pattern
-                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d{5})(?:[^a-zA-Z0-9]|\\Z)" );
-    private final static Pattern GENBANK_NUCLEOTIDE_AC_PATTERN_2 = Pattern
-                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}\\d{6})(?:[^a-zA-Z0-9]|\\Z)" );
-    private final static Pattern GENBANK_PROTEIN_AC_PATTERN      = Pattern
-                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5})(?:[^a-zA-Z0-9]|\\Z)" );
-    private final static boolean DEBUG                           = false;
-
-    /**
-     * Returns null if no match.
-     * 
-     * @param query
-     * @param db 
-     * @return
-     */
-    static public String parseGenbankAccessor( final String query ) {
-        Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );
-        if ( m.lookingAt() ) {
-            return m.group( 1 );
-        }
-        else {
-            m = GENBANK_NUCLEOTIDE_AC_PATTERN_2.matcher( query );
-            if ( m.lookingAt() ) {
-                return m.group( 1 );
-            }
-            else {
-                m = GENBANK_PROTEIN_AC_PATTERN.matcher( query );
-                if ( m.lookingAt() ) {
-                    return m.group( 1 );
-                }
-                else {
-                    return null;
-                }
-            }
-        }
-    }
-
+   
     static String extract( final String target, final String a, final String b ) {
         final int i_a = target.indexOf( a );
         final int i_b = target.indexOf( b );