X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Futil%2FSequenceAccessionTools.java;h=cce1b2e578d469863cce64fe343f6fbe2623ceb3;hb=a1114eb8610e592961a40e5c3d46d647c02b5108;hp=4136049ad6ff6eee146ffc8c7db42ee67c1afce4;hpb=3171b9b28762e02b3fbd9b5f8a5b1946497f2178;p=jalview.git

diff --git a/forester/java/src/org/forester/util/SequenceAccessionTools.java b/forester/java/src/org/forester/util/SequenceAccessionTools.java
index 4136049..cce1b2e 100644
--- a/forester/java/src/org/forester/util/SequenceAccessionTools.java
+++ b/forester/java/src/org/forester/util/SequenceAccessionTools.java
@@ -33,16 +33,11 @@ import java.util.regex.Pattern;
 
 import org.forester.phylogeny.PhylogenyNode;
 import org.forester.phylogeny.data.Accession;
+import org.forester.phylogeny.data.Accession.Source;
 import org.forester.phylogeny.data.Sequence;
 
 public final class SequenceAccessionTools {
 
-    public final static Pattern  UNIPROT_KB_PATTERN_0            = Pattern
-                                                                         .compile( "\\b([A-Z][0-9][A-Z0-9]{3}[0-9])\\b" );
-    public final static Pattern  UNIPROT_KB_PATTERN_1            = Pattern
-                                                                         .compile( "(?:\\b|_)(?:sp|tr)[\\.|\\-_=/\\\\]([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );
-    public final static Pattern  UNIPROT_KB_PATTERN_2            = Pattern
-                                                                         .compile( "(?:\\b|_)(?:[A-Z0-9]{2,5}|(?:[A-Z][0-9][A-Z0-9]{3}[0-9]))_(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA)(?:\\b|_)" );
     // gb_ADF31344_1_segmented_worms_
     // gb_AAA96518_1
     // gb_EHB07727_1_rodents_
@@ -56,230 +51,214 @@ public final class SequenceAccessionTools {
     //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}(?:\\.\\d+)?)(?:[^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}(?:\\.\\d+)?)(?:[^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}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );
-    private final static Pattern GI_PATTERN                      = Pattern
-                                                                         .compile( "(?:\\b|_)(?:GI|gi)[|_=:](\\d+)(?:\\b|_)" );
+    public final static Pattern  GENBANK_NUC_PATTERN_1 = Pattern
+                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );
+    public final static Pattern  GENBANK_NUC_PATTERN_2 = Pattern
+                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}\\d{6}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );
+    public final static Pattern  GENBANK_PROT_PATTERN  = Pattern
+                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );
+    public final static Pattern  GI_PATTERN            = Pattern.compile( "(?:\\b|_)(?:GI|gi)[|_=:](\\d+)(?:\\b|_)" );
+    public final static Pattern  UNIPROT_KB_PATTERN_0  = Pattern
+                                                               .compile( "(?:\\b|_)([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );
+    public final static Pattern  UNIPROT_KB_PATTERN_1  = Pattern
+                                                               .compile( "(?:\\b|_)(?:sp|tr)[\\.|\\-_=/\\\\]([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );
+    public final static Pattern  UNIPROT_KB_PATTERN_2  = Pattern
+                                                               .compile( "(?:\\b|_)(?:[A-Z0-9]{2,5}|(?:[A-Z][0-9][A-Z0-9]{3}[0-9]))_(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA)(?:\\b|_)" );
     // 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)" );
+    private final static Pattern REFSEQ_PATTERN        = Pattern
+                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}_\\d{6,})(?:[^a-zA-Z0-9]|\\Z)" );
 
     private SequenceAccessionTools() {
         // Hiding the constructor.
     }
 
-    public static String extractGenbankAccessor( final PhylogenyNode node ) {
-        String v = null;
-        if ( node.getNodeData().isHasSequence() ) {
-            final Sequence seq = node.getNodeData().getSequence();
-            if ( !ForesterUtil.isEmpty( seq.getSymbol() ) ) {
-                v = parseGenbankAccessor( seq.getSymbol() );
-            }
-            if ( !ForesterUtil.isEmpty( seq.getGeneName() ) ) {
-                v = parseGenbankAccessor( seq.getGeneName() );
-            }
-            if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
-                v = parseGenbankAccessor( seq.getName() );
-            }
-            if ( ForesterUtil.isEmpty( v ) && ( node.getNodeData().getSequence().getAccession() != null )
-                    && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
-                v = parseGenbankAccessor( seq.getAccession().getValue() );
-            }
+    public final static boolean isProteinDbQuery( final String query ) {
+        final String r1 = parseRefSeqAccessorFromString( query );
+        if ( !ForesterUtil.isEmpty( r1 ) && ( r1.charAt( 1 ) == 'P' ) ) {
+            return true;
+        }
+        final String r2 = parseUniProtAccessorFromString( query );
+        if ( !ForesterUtil.isEmpty( r2 ) ) {
+            return true;
         }
-        if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( node.getName() ) ) {
-            v = parseGenbankAccessor( node.getName() );
+        return GENBANK_PROT_PATTERN.matcher( query ).lookingAt();
+    }
+
+    public final static Accession obtainAccessorFromDataFields( final PhylogenyNode n ) {
+        String a = obtainUniProtAccessorFromDataFields( n );
+        if ( !ForesterUtil.isEmpty( a ) ) {
+            return new Accession( a, Source.UNIPROT );
+        }
+        a = obtainGenbankAccessorFromDataFields( n );
+        if ( !ForesterUtil.isEmpty( a ) ) {
+            return new Accession( a, Source.NCBI );
         }
-        return v;
+        a = obtainRefSeqAccessorFromDataFields( n );
+        if ( !ForesterUtil.isEmpty( a ) ) {
+            return new Accession( a, Source.REFSEQ );
+        }
+        a = obtainGiNumberFromDataFields( n );
+        if ( !ForesterUtil.isEmpty( a ) ) {
+            return new Accession( a, Source.GI );
+        }
+        return null;
     }
 
-    public static String extractGInumber( final PhylogenyNode node ) {
-        String v = null;
-        if ( node.getNodeData().isHasSequence() ) {
-            final Sequence seq = node.getNodeData().getSequence();
-            if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
-                v = parseGInumber( seq.getName() );
+    public final static Accession obtainFromSeqAccession( final PhylogenyNode n ) {
+        if ( n.getNodeData().isHasSequence() && ( n.getNodeData().getSequence().getAccession() != null )
+                && !ForesterUtil.isEmpty( n.getNodeData().getSequence().getAccession().getSource() )
+                && !ForesterUtil.isEmpty( n.getNodeData().getSequence().getAccession().getValue() ) ) {
+            final String source = n.getNodeData().getSequence().getAccession().getSource().toLowerCase();
+            final String value = n.getNodeData().getSequence().getAccession().getValue();
+            if ( ( source.startsWith( "uniprot" ) || source.equals( "swissprot" ) || source.equals( "trembl" ) || source
+                    .equals( "sp" ) ) ) {
+                return new Accession( value, Source.UNIPROT );
             }
-            if ( ForesterUtil.isEmpty( v ) && ( node.getNodeData().getSequence().getAccession() != null )
-                    && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
-                v = parseGInumber( seq.getAccession().getValue() );
+            else if ( source.equals( "embl" ) || source.equals( "ebi" ) ) {
+                return new Accession( value, Source.EMBL );
+            }
+            else if ( source.equals( "ncbi" ) || source.equals( "genbank" ) ) {
+                return new Accession( value, Source.NCBI );
+            }
+            else if ( source.equals( "refseq" ) ) {
+                return new Accession( value, Source.REFSEQ );
+            }
+            else if ( source.equals( "gi" ) ) {
+                return new Accession( value, Source.GI );
             }
         }
-        if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( node.getName() ) ) {
-            v = parseGInumber( node.getName() );
-        }
-        return v;
+        return null;
     }
 
-    public static String extractRefSeqAccessor( final PhylogenyNode node ) {
-        String v = null;
-        if ( node.getNodeData().isHasSequence() ) {
-            final Sequence seq = node.getNodeData().getSequence();
+    public final static String obtainGenbankAccessorFromDataFields( final PhylogenyNode n ) {
+        String a = null;
+        if ( n.getNodeData().isHasSequence() ) {
+            final Sequence seq = n.getNodeData().getSequence();
             if ( !ForesterUtil.isEmpty( seq.getSymbol() ) ) {
-                v = parseRefSeqAccessor( seq.getSymbol() );
+                a = parseGenbankAccessorFromString( seq.getSymbol() );
             }
             if ( !ForesterUtil.isEmpty( seq.getGeneName() ) ) {
-                v = parseRefSeqAccessor( seq.getGeneName() );
+                a = parseGenbankAccessorFromString( seq.getGeneName() );
             }
-            if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
-                v = parseRefSeqAccessor( seq.getName() );
+            if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
+                a = parseGenbankAccessorFromString( seq.getName() );
             }
-            if ( ForesterUtil.isEmpty( v ) && ( node.getNodeData().getSequence().getAccession() != null )
+            if ( ForesterUtil.isEmpty( a ) && ( n.getNodeData().getSequence().getAccession() != null )
                     && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
-                v = parseRefSeqAccessor( seq.getAccession().getValue() );
+                a = parseGenbankAccessorFromString( seq.getAccession().getValue() );
             }
         }
-        if ( ForesterUtil.isEmpty( v ) && !ForesterUtil.isEmpty( node.getName() ) ) {
-            v = parseRefSeqAccessor( node.getName() );
+        if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( n.getName() ) ) {
+            a = parseGenbankAccessorFromString( n.getName() );
         }
-        return v;
+        return a;
     }
 
-    public static String extractUniProtKbProteinSeqIdentifier( final PhylogenyNode node ) {
+    public final static String obtainGiNumberFromDataFields( final PhylogenyNode n ) {
         String a = null;
-        if ( node.getNodeData().isHasSequence() ) {
-            final Sequence seq = node.getNodeData().getSequence();
-            if ( !ForesterUtil.isEmpty( seq.getSymbol() ) ) {
-                a = SequenceAccessionTools.extractUniProtKbProteinSeqIdentifier( seq.getSymbol() );
-            }
+        if ( n.getNodeData().isHasSequence() ) {
+            final Sequence seq = n.getNodeData().getSequence();
             if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
-                a = SequenceAccessionTools.extractUniProtKbProteinSeqIdentifier( seq.getName() );
+                a = parseGInumberFromString( seq.getName() );
             }
             if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getGeneName() ) ) {
-                a = SequenceAccessionTools.extractUniProtKbProteinSeqIdentifier( seq.getGeneName() );
+                a = parseGInumberFromString( seq.getGeneName() );
             }
-            if ( ForesterUtil.isEmpty( a ) && ( node.getNodeData().getSequence().getAccession() != null )
+            if ( ForesterUtil.isEmpty( a ) && ( n.getNodeData().getSequence().getAccession() != null )
                     && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
-                a = SequenceAccessionTools.extractUniProtKbProteinSeqIdentifier( seq.getAccession().getValue() );
+                a = parseGInumberFromString( seq.getAccession().getValue() );
             }
         }
-        if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( node.getName() ) ) {
-            a = SequenceAccessionTools.extractUniProtKbProteinSeqIdentifier( node.getName() );
+        if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( n.getName() ) ) {
+            a = parseGInumberFromString( n.getName() );
         }
         return a;
     }
 
-    public static String extractUniProtKbProteinSeqIdentifier( final String str ) {
-        Matcher m = UNIPROT_KB_PATTERN_0.matcher( str );
-        if ( m.find() ) {
-            return m.group( 1 );
-        }
-        m = UNIPROT_KB_PATTERN_1.matcher( str );
-        if ( m.find() ) {
-            return m.group( 1 );
-        }
-        m = UNIPROT_KB_PATTERN_2.matcher( str );
-        if ( m.find() ) {
-            return m.group();
-        }
-        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 = extractUniProtKbProteinSeqIdentifier( query );
-        if ( !ForesterUtil.isEmpty( r2 ) ) {
-            return true;
-        }
-        return GENBANK_PROTEIN_AC_PATTERN.matcher( query ).lookingAt();
-    }
-
-    public final static Accession parse( final PhylogenyNode n ) {
-        String v = extractUniProtKbProteinSeqIdentifier( n );
-        if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Accession( v, Accession.UNIPROT );
-        }
-        v = extractGenbankAccessor( n );
-        if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Accession( v, Accession.NCBI );
-        }
-        v = extractRefSeqAccessor( n );
-        if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Accession( v, Accession.REFSEQ );
+    public final static String obtainRefSeqAccessorFromDataFields( final PhylogenyNode n ) {
+        String a = null;
+        if ( n.getNodeData().isHasSequence() ) {
+            final Sequence seq = n.getNodeData().getSequence();
+            if ( !ForesterUtil.isEmpty( seq.getSymbol() ) ) {
+                a = parseRefSeqAccessorFromString( seq.getSymbol() );
+            }
+            if ( !ForesterUtil.isEmpty( seq.getGeneName() ) ) {
+                a = parseRefSeqAccessorFromString( seq.getGeneName() );
+            }
+            if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
+                a = parseRefSeqAccessorFromString( seq.getName() );
+            }
+            if ( ForesterUtil.isEmpty( a ) && ( n.getNodeData().getSequence().getAccession() != null )
+                    && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
+                a = parseRefSeqAccessorFromString( seq.getAccession().getValue() );
+            }
         }
-        v = extractGInumber( n );
-        if ( !ForesterUtil.isEmpty( v ) ) {
-            return new Accession( v, Accession.GI );
+        if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( n.getName() ) ) {
+            a = parseRefSeqAccessorFromString( n.getName() );
         }
-        return null;
+        return a;
     }
 
-    public final static Accession obtainFromSeqAccession( final PhylogenyNode node ) {
-        if ( node.getNodeData().isHasSequence() && ( node.getNodeData().getSequence().getAccession() != null )
-                && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
-                && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() ) ) {
-            final String source = node.getNodeData().getSequence().getAccession().getSource().toLowerCase();
-            final String value = node.getNodeData().getSequence().getAccession().getValue();
-            if ( ( source.startsWith( "uniprot" ) || source.equals( "swissprot" ) || source.equals( "trembl" ) || source
-                    .equals( "sp" ) ) ) {
-                return new Accession( value, Accession.UNIPROT );
-            }
-            else if ( source.equals( "embl" ) || source.equals( "ebi" ) ) {
-                return new Accession( value, Accession.EMBL );
+    public final static String obtainUniProtAccessorFromDataFields( final PhylogenyNode n ) {
+        String a = null;
+        if ( n.getNodeData().isHasSequence() ) {
+            final Sequence seq = n.getNodeData().getSequence();
+            if ( !ForesterUtil.isEmpty( seq.getSymbol() ) ) {
+                a = SequenceAccessionTools.parseUniProtAccessorFromString( seq.getSymbol() );
             }
-            else if ( source.equals( "ncbi" ) || source.equals( "genbank" ) ) {
-                return new Accession( value, Accession.NCBI );
+            if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getName() ) ) {
+                a = SequenceAccessionTools.parseUniProtAccessorFromString( seq.getName() );
             }
-            else if ( source.equals( "refseq" ) ) {
-                return new Accession( value, Accession.REFSEQ );
+            if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( seq.getGeneName() ) ) {
+                a = SequenceAccessionTools.parseUniProtAccessorFromString( seq.getGeneName() );
             }
-            else if ( source.equals( "gi" ) ) {
-                return new Accession( value, Accession.GI );
+            if ( ForesterUtil.isEmpty( a ) && ( n.getNodeData().getSequence().getAccession() != null )
+                    && !ForesterUtil.isEmpty( seq.getAccession().getValue() ) ) {
+                a = SequenceAccessionTools.parseUniProtAccessorFromString( seq.getAccession().getValue() );
             }
         }
-        return null;
+        if ( ForesterUtil.isEmpty( a ) && !ForesterUtil.isEmpty( n.getName() ) ) {
+            a = SequenceAccessionTools.parseUniProtAccessorFromString( n.getName() );
+        }
+        return a;
     }
 
-    /**
-     * Returns null if no match.
-     * 
-     */
-    public final static Accession parse( final String s ) {
+    public final static Accession parseAccessorFromString( final String s ) {
         if ( !ForesterUtil.isEmpty( s ) ) {
-            String v = extractUniProtKbProteinSeqIdentifier( s );
+            String v = parseUniProtAccessorFromString( s );
             if ( !ForesterUtil.isEmpty( v ) ) {
-                return new Accession( v, Accession.UNIPROT );
+                return new Accession( v, Source.UNIPROT );
             }
-            v = parseGenbankAccessor( s );
+            v = parseGenbankAccessorFromString( s );
             if ( !ForesterUtil.isEmpty( v ) ) {
-                return new Accession( v, Accession.NCBI );
+                return new Accession( v, Source.NCBI );
             }
-            v = parseRefSeqAccessor( s );
+            v = parseRefSeqAccessorFromString( s );
             if ( !ForesterUtil.isEmpty( v ) ) {
-                return new Accession( v, Accession.REFSEQ );
+                return new Accession( v, Source.REFSEQ );
             }
-            v = parseGInumber( s );
+            v = parseGInumberFromString( s );
             if ( !ForesterUtil.isEmpty( v ) ) {
-                return new Accession( v, Accession.GI );
+                return new Accession( v, Source.GI );
             }
         }
         return null;
     }
 
-    /**
-     * Returns null if no match.
-     * 
-     */
-    public static String parseGenbankAccessor( final String query ) {
-        Matcher m = GENBANK_NUCLEOTIDE_AC_PATTERN_1.matcher( query );
+    public final static String parseGenbankAccessorFromString( final String s ) {
+        Matcher m = GENBANK_NUC_PATTERN_1.matcher( s );
         if ( m.lookingAt() ) {
             return m.group( 1 );
         }
         else {
-            m = GENBANK_NUCLEOTIDE_AC_PATTERN_2.matcher( query );
+            m = GENBANK_NUC_PATTERN_2.matcher( s );
             if ( m.lookingAt() ) {
                 return m.group( 1 );
             }
             else {
-                m = GENBANK_PROTEIN_AC_PATTERN.matcher( query );
+                m = GENBANK_PROT_PATTERN.matcher( s );
                 if ( m.lookingAt() ) {
                     return m.group( 1 );
                 }
@@ -290,8 +269,8 @@ public final class SequenceAccessionTools {
         }
     }
 
-    public static String parseGenbankProteinAccessor( final String query ) {
-        final Matcher m = GENBANK_PROTEIN_AC_PATTERN.matcher( query );
+    public final static String parseGenbankProteinAccessorFromString( final String s ) {
+        final Matcher m = GENBANK_PROT_PATTERN.matcher( s );
         if ( m.lookingAt() ) {
             return m.group( 1 );
         }
@@ -300,23 +279,35 @@ public final class SequenceAccessionTools {
         }
     }
 
-    public static String parseGInumber( final String query ) {
-        final Matcher m = GI_PATTERN.matcher( query );
+    public final static String parseGInumberFromString( final String s ) {
+        final Matcher m = GI_PATTERN.matcher( s );
         if ( m.find() ) {
             return m.group( 1 );
         }
         return null;
     }
 
-    /**
-     * Returns null if no match.
-     * 
-     */
-    public final static String parseRefSeqAccessor( final String query ) {
-        final Matcher m = REFSEQ_PATTERN.matcher( query );
+    public final static String parseRefSeqAccessorFromString( final String s ) {
+        final Matcher m = REFSEQ_PATTERN.matcher( s );
         if ( m.lookingAt() ) {
             return m.group( 1 );
         }
         return null;
     }
+
+    public final static String parseUniProtAccessorFromString( final String s ) {
+        Matcher m = UNIPROT_KB_PATTERN_1.matcher( s );
+        if ( m.find() ) {
+            return m.group( 1 );
+        }
+        m = UNIPROT_KB_PATTERN_2.matcher( s );
+        if ( m.find() ) {
+            return m.group();
+        }
+        m = UNIPROT_KB_PATTERN_0.matcher( s );
+        if ( m.find() ) {
+            return m.group( 1 );
+        }
+        return null;
+    }
 }