X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Futil%2FSequenceIdParser.java;h=3b6bc5d0acd7ad902881db58d48644bf5da56557;hb=03e51d179caedf757b09e2872f9500318bd85a53;hp=8def26026d0fbecc168a4d847cea97205a2ba237;hpb=34e555e6895d06fa5b7de6a646c37b0817efafc6;p=jalview.git

diff --git a/forester/java/src/org/forester/util/SequenceIdParser.java b/forester/java/src/org/forester/util/SequenceIdParser.java
index 8def260..3b6bc5d 100644
--- a/forester/java/src/org/forester/util/SequenceIdParser.java
+++ b/forester/java/src/org/forester/util/SequenceIdParser.java
@@ -32,12 +32,9 @@ import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
 import org.forester.phylogeny.data.Identifier;
-import org.forester.ws.uniprot.DatabaseTools;
 
 public final class SequenceIdParser {
 
-    
-   
     // gb_ADF31344_1_segmented_worms_
     // gb_AAA96518_1
     // gb_EHB07727_1_rodents_
@@ -47,7 +44,6 @@ public final class SequenceIdParser {
     // mites|ref_XP_002434188_1
     // ref_XP_002434188_1_mites___ticks_
     // ref_NP_001121530_1_frogs___toads_
-    
     //The format for GenBank Accession numbers are:
     //Nucleotide: 1 letter + 5 numerals OR 2 letters + 6 numerals
     //Protein:    3 letters + 5 numerals
@@ -58,9 +54,15 @@ public final class SequenceIdParser {
                                                                          .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                           = true;
+    // RefSeq accession numbers can be distinguished from GenBank accessions 
+    // by their distinct prefix format of 2 characters followed by an
+    // underscore character ('_'). For example, a RefSeq protein accession is NP_015325. 
+    private final static Pattern REFSEQ_PATTERN                  = Pattern
+                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}_\\d{6,})(?:[^a-zA-Z0-9]|\\Z)" );
+    // See: http://web.expasy.org/docs/userman.html#ID_line
+    private final static Pattern TREMBL_PATTERN                  = Pattern
+                                                                         .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z][0-9][A-Z0-9]{3}[0-9])(?:[^a-zA-Z0-9]|\\Z)" );
 
-    
     /**
      * Returns null if no match.
      * 
@@ -68,23 +70,36 @@ public final class SequenceIdParser {
     public final static Identifier parse( final String s ) {
         String v = parseGenbankAccessor( s );
         if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Identifier( v, "ncbi" );
+            return new Identifier( v, Identifier.NCBI );
         }
         v = parseRefSeqAccessor( s );
         if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Identifier( v, "ncbi" );
+            return new Identifier( v, Identifier.REFSEQ );
+        }
+        v = parseTrEMBLAccessor( s );
+        if ( !ForesterUtil.isEmpty( v ) ) {
+            return new Identifier( v, Identifier.SP );
         }
         return null;
     }
-    
+
+    public final static boolean isProtein( final String query ) {
+        final String r1 = parseRefSeqAccessor( query );
+        if ( !ForesterUtil.isEmpty( r1 ) && ( r1.charAt( 1 ) == 'P' ) ) {
+            return true;
+        }
+        final String r2 = parseTrEMBLAccessor( query );
+        if ( !ForesterUtil.isEmpty( r2 ) ) {
+            return true;
+        }
+        return GENBANK_PROTEIN_AC_PATTERN.matcher( query ).lookingAt();
+    }
+
     /**
      * Returns null if no match.
      * 
-     * @param query
-     * @param db 
-     * @return
      */
-    static public String parseGenbankAccessor( final String query ) {
+    public static String parseGenbankAccessor( final String query ) {
         Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );
         if ( m.lookingAt() ) {
             return m.group( 1 );
@@ -105,35 +120,32 @@ public final class SequenceIdParser {
             }
         }
     }
-    
-    public final static String parseRefSeqAccessor( final String query ) {
-        Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );
+
+    /**
+     * Returns null if no match.
+     * 
+     */
+    private final static String parseRefSeqAccessor( final String query ) {
+        final Matcher m = REFSEQ_PATTERN.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;
-                }
-            }
+        return null;
+    }
+
+    /**
+     * Returns null if no match.
+     * 
+     */
+    private final static String parseTrEMBLAccessor( final String query ) {
+        final Matcher m = TREMBL_PATTERN.matcher( query );
+        if ( m.lookingAt() ) {
+            return m.group( 1 );
         }
+        return null;
     }
-    
-    
-    
+
     private SequenceIdParser() {
         // Hiding the constructor.
     }
-    
-    
-    
 }