inprogress

[jalview.git] / forester / java / src / org / forester / util / SequenceAccessionTools.java

diff --git a/forester/java/src/org/forester/util/SequenceAccessionTools.java b/forester/java/src/org/forester/util/SequenceAccessionTools.java
index c65fbc5..cce1b2e 100644 (file)
--- a/forester/java/src/org/forester/util/SequenceAccessionTools.java
+++ b/forester/java/src/org/forester/util/SequenceAccessionTools.java
@@ -33,6 +33,7 @@ import java.util.regex.Pattern;
 \r
 import org.forester.phylogeny.PhylogenyNode;\r
 import org.forester.phylogeny.data.Accession;\r
+import org.forester.phylogeny.data.Accession.Source;\r
 import org.forester.phylogeny.data.Sequence;\r
 \r
 public final class SequenceAccessionTools {\r
@@ -57,7 +58,8 @@ public final class SequenceAccessionTools {
     public final static Pattern  GENBANK_PROT_PATTERN  = Pattern\r
                                                                .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
     public final static Pattern  GI_PATTERN            = Pattern.compile( "(?:\\b|_)(?:GI|gi)[|_=:](\\d+)(?:\\b|_)" );\r
-    public final static Pattern  UNIPROT_KB_PATTERN_0  = Pattern.compile( "\\b([A-Z][0-9][A-Z0-9]{3}[0-9])\\b" );\r
+    public final static Pattern  UNIPROT_KB_PATTERN_0  = Pattern\r
+                                                               .compile( "(?:\\b|_)([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );\r
     public final static Pattern  UNIPROT_KB_PATTERN_1  = Pattern\r
                                                                .compile( "(?:\\b|_)(?:sp|tr)[\\.|\\-_=/\\\\]([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );\r
     public final static Pattern  UNIPROT_KB_PATTERN_2  = Pattern\r
@@ -87,19 +89,19 @@ public final class SequenceAccessionTools {
     public final static Accession obtainAccessorFromDataFields( final PhylogenyNode n ) {\r
         String a = obtainUniProtAccessorFromDataFields( n );\r
         if ( !ForesterUtil.isEmpty( a ) ) {\r
-            return new Accession( a, Accession.UNIPROT );\r
+            return new Accession( a, Source.UNIPROT );\r
         }\r
         a = obtainGenbankAccessorFromDataFields( n );\r
         if ( !ForesterUtil.isEmpty( a ) ) {\r
-            return new Accession( a, Accession.NCBI );\r
+            return new Accession( a, Source.NCBI );\r
         }\r
         a = obtainRefSeqAccessorFromDataFields( n );\r
         if ( !ForesterUtil.isEmpty( a ) ) {\r
-            return new Accession( a, Accession.REFSEQ );\r
+            return new Accession( a, Source.REFSEQ );\r
         }\r
         a = obtainGiNumberFromDataFields( n );\r
         if ( !ForesterUtil.isEmpty( a ) ) {\r
-            return new Accession( a, Accession.GI );\r
+            return new Accession( a, Source.GI );\r
         }\r
         return null;\r
     }\r
@@ -112,19 +114,19 @@ public final class SequenceAccessionTools {
             final String value = n.getNodeData().getSequence().getAccession().getValue();\r
             if ( ( source.startsWith( "uniprot" ) || source.equals( "swissprot" ) || source.equals( "trembl" ) || source\r
                     .equals( "sp" ) ) ) {\r
-                return new Accession( value, Accession.UNIPROT );\r
+                return new Accession( value, Source.UNIPROT );\r
             }\r
             else if ( source.equals( "embl" ) || source.equals( "ebi" ) ) {\r
-                return new Accession( value, Accession.EMBL );\r
+                return new Accession( value, Source.EMBL );\r
             }\r
             else if ( source.equals( "ncbi" ) || source.equals( "genbank" ) ) {\r
-                return new Accession( value, Accession.NCBI );\r
+                return new Accession( value, Source.NCBI );\r
             }\r
             else if ( source.equals( "refseq" ) ) {\r
-                return new Accession( value, Accession.REFSEQ );\r
+                return new Accession( value, Source.REFSEQ );\r
             }\r
             else if ( source.equals( "gi" ) ) {\r
-                return new Accession( value, Accession.GI );\r
+                return new Accession( value, Source.GI );\r
             }\r
         }\r
         return null;\r
@@ -227,19 +229,19 @@ public final class SequenceAccessionTools {
         if ( !ForesterUtil.isEmpty( s ) ) {\r
             String v = parseUniProtAccessorFromString( s );\r
             if ( !ForesterUtil.isEmpty( v ) ) {\r
-                return new Accession( v, Accession.UNIPROT );\r
+                return new Accession( v, Source.UNIPROT );\r
             }\r
             v = parseGenbankAccessorFromString( s );\r
             if ( !ForesterUtil.isEmpty( v ) ) {\r
-                return new Accession( v, Accession.NCBI );\r
+                return new Accession( v, Source.NCBI );\r
             }\r
             v = parseRefSeqAccessorFromString( s );\r
             if ( !ForesterUtil.isEmpty( v ) ) {\r
-                return new Accession( v, Accession.REFSEQ );\r
+                return new Accession( v, Source.REFSEQ );\r
             }\r
             v = parseGInumberFromString( s );\r
             if ( !ForesterUtil.isEmpty( v ) ) {\r
-                return new Accession( v, Accession.GI );\r
+                return new Accession( v, Source.GI );\r
             }\r
         }\r
         return null;\r
@@ -294,11 +296,7 @@ public final class SequenceAccessionTools {
     }\r
 \r
     public final static String parseUniProtAccessorFromString( final String s ) {\r
-        Matcher m = UNIPROT_KB_PATTERN_0.matcher( s );\r
-        if ( m.find() ) {\r
-            return m.group( 1 );\r
-        }\r
-        m = UNIPROT_KB_PATTERN_1.matcher( s );\r
+        Matcher m = UNIPROT_KB_PATTERN_1.matcher( s );\r
         if ( m.find() ) {\r
             return m.group( 1 );\r
         }\r
@@ -306,6 +304,10 @@ public final class SequenceAccessionTools {
         if ( m.find() ) {\r
             return m.group();\r
         }\r
+        m = UNIPROT_KB_PATTERN_0.matcher( s );\r
+        if ( m.find() ) {\r
+            return m.group( 1 );\r
+        }\r
         return null;\r
     }\r
 }\r