git://source.jalview.org / jalview.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
JAL-2844 work started on moving the graphical partitioning code to Aptx
[jalview.git] / forester / java / src / org / forester / ws / seqdb / UniProtEntry.java
diff --git a/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java b/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
index 77ebe4e..ba2d7a4 100644 (file)
--- a/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
+++ b/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
@@ -42,18 +42,19 @@ import org.forester.util.ForesterUtil;
 
 public final class UniProtEntry implements SequenceDatabaseEntry {
 
-    public final static Pattern  BindingDB_PATTERN = Pattern.compile( "BindingDB;\\s+([0-9A-Z]+);" );
-    public final static Pattern  CTD_PATTERN       = Pattern.compile( "CTD;\\s+(\\d+);" );
-    public final static Pattern  DrugBank_PATTERN  = Pattern.compile( "DrugBank;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
-    public final static Pattern  GO_PATTERN        = Pattern.compile( "GO;\\s+(GO:\\d+);\\s+([PFC]):([^;]+);" );
-    public final static Pattern  KEGG_PATTERN      = Pattern.compile( "KEGG;\\s+([a-z]+:[0-9]+);" );
-    public final static Pattern  MIM_PATTERN       = Pattern.compile( "MIM;\\s+(\\d+);" );
-    public final static Pattern  NextBio_PATTERN   = Pattern.compile( "NextBio;\\s+(\\d+);" );
-    public final static Pattern  Orphanet_PATTERN  = Pattern.compile( "Orphanet;\\s+(\\d+);\\s+([^\\.]+)" );
-    public final static Pattern  PDB_PATTERN       = Pattern.compile( "PDB;\\s+([0-9A-Z]{4});\\s+([^;]+)" );
-    public final static Pattern  PharmGKB_PATTERN  = Pattern.compile( "PharmGKB;\\s+([0-9A-Z]+);" );
-    public final static Pattern  Reactome_PATTERN  = Pattern.compile( "Reactome;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
-    public final static Pattern  HGNC_PATTERN      = Pattern.compile( "HGNC;\\s+HGNC:(\\d+);" );
+    public final static Pattern  BindingDB_PATTERN  = Pattern.compile( "BindingDB;\\s+([0-9A-Z]+);" );
+    public final static Pattern  CTD_PATTERN        = Pattern.compile( "CTD;\\s+(\\d+);" );
+    public final static Pattern  DrugBank_PATTERN   = Pattern.compile( "DrugBank;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
+    public final static Pattern  GO_PATTERN         = Pattern.compile( "GO;\\s+(GO:\\d+);\\s+([PFC]):([^;]+);" );
+    public final static Pattern  KEGG_PATTERN       = Pattern.compile( "KEGG;\\s+([a-z]+:[0-9]+);" );
+    public final static Pattern  MIM_PATTERN        = Pattern.compile( "MIM;\\s+(\\d+);" );
+    public final static Pattern  NextBio_PATTERN    = Pattern.compile( "NextBio;\\s+(\\d+);" );
+    public final static Pattern  Orphanet_PATTERN   = Pattern.compile( "Orphanet;\\s+(\\d+);\\s+([^\\.]+)" );
+    public final static Pattern  PDB_PATTERN        = Pattern.compile( "PDB;\\s+([0-9A-Z]{4});\\s+([^;]+)" );
+    public final static Pattern  PharmGKB_PATTERN   = Pattern.compile( "PharmGKB;\\s+([0-9A-Z]+);" );
+    public final static Pattern  Reactome_PATTERN   = Pattern.compile( "Reactome;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
+    public final static Pattern  HGNC_PATTERN       = Pattern.compile( "HGNC;\\s+HGNC:(\\d+);" );
+    public final static Pattern  NCBI_TAXID_PATTERN = Pattern.compile( "NCBI_TaxID=(\\d+)" );
     private String               _ac;
     private SortedSet<Accession> _cross_references;
     private String               _gene_name;
@@ -126,6 +127,21 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
                 .isEmpty() ) );
     }
 
+    @Override
+    public String getMap() {
+        return null;
+    }
+
+    @Override
+    public String getChromosome() {
+        return null;
+    }
+
+    @Override
+    public MolecularSequence getMolecularSequence() {
+        return _mol_seq;
+    }
+    
     private void addCrossReference( final Accession accession ) {
         if ( _cross_references == null ) {
             _cross_references = new TreeSet<Accession>();
@@ -178,32 +194,58 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
         }
     }
 
-    public static SequenceDatabaseEntry createInstanceFromPlainText( final List<String> lines ) {
+   
+
+    @Override
+    public SortedSet<Annotation> getAnnotations() {
+        return null;
+    }
+
+    public final static SequenceDatabaseEntry createInstance( final List<String> lines ) {
         final UniProtEntry e = new UniProtEntry();
         boolean saw_sq = false;
         final StringBuffer sq_buffer = new StringBuffer();
         boolean is_aa = false;
         for( final String line : lines ) {
-            //System.out.println( line );
             if ( line.startsWith( "AC" ) ) {
                 e.setAc( SequenceDbWsTools.extractFromTo( line, "AC", ";" ) );
             }
             else if ( line.startsWith( "DE" ) && ForesterUtil.isEmpty( e.getSequenceName() ) ) {
                 if ( ( line.indexOf( "RecName:" ) > 0 ) && ( line.indexOf( "Full=" ) > 0 ) ) {
-                    e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
+                    if ( line.indexOf( "{" ) > 0 ) {
+                        e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", "{" ) );
+                    }
+                    else {
+                        e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
+                    }
                 }
                 else if ( ( line.indexOf( "SubName:" ) > 0 ) && ( line.indexOf( "Full=" ) > 0 ) ) {
-                    e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
+                    if ( line.indexOf( "{" ) > 0 ) {
+                        e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", "{" ) );
+                    }
+                    else {
+                        e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
+                    }
                 }
             }
             else if ( line.startsWith( "DE" ) && ForesterUtil.isEmpty( e.getSequenceSymbol() ) ) {
                 if ( line.indexOf( "Short=" ) > 0 ) {
-                    e.setSequenceSymbol( SequenceDbWsTools.extractFromTo( line, "Short=", ";" ) );
+                    if ( line.indexOf( "{" ) > 0 ) {
+                        e.setSequenceSymbol( SequenceDbWsTools.extractFromTo( line, "Short=", "{" ) );
+                    }
+                    else {
+                        e.setSequenceSymbol( SequenceDbWsTools.extractFromTo( line, "Short=", ";" ) );
+                    }
                 }
             }
             else if ( line.startsWith( "GN" ) && ForesterUtil.isEmpty( e.getGeneName() ) ) {
                 if ( line.indexOf( "Name=" ) > 0 ) {
-                    e.setGeneName( SequenceDbWsTools.extractFromTo( line, "Name=", ";" ) );
+                    if ( line.indexOf( "{" ) > 0 ) {
+                        e.setGeneName( SequenceDbWsTools.extractFromTo( line, "Name=", "{" ) );
+                    }
+                    else {
+                        e.setGeneName( SequenceDbWsTools.extractFromTo( line, "Name=", ";" ) );
+                    }
                 }
             }
             else if ( line.startsWith( "DR" ) ) {
@@ -300,7 +342,10 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
             }
             else if ( line.startsWith( "OX" ) ) {
                 if ( line.indexOf( "NCBI_TaxID=" ) > 0 ) {
-                    e.setTaxId( SequenceDbWsTools.extractFromTo( line, "NCBI_TaxID=", ";" ) );
+                    final Matcher m = NCBI_TAXID_PATTERN.matcher( line );
+                    if ( m.find() ) {
+                        e.setTaxId( m.group( 1 ) );
+                    }
                 }
             }
             else if ( line.startsWith( "SQ" ) ) {
@@ -313,29 +358,14 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
                 sq_buffer.append( line.replaceAll( "\\s+", "" ) );
             }
         }
-        if ( ( sq_buffer.length() > 0 ) && is_aa ) {
-            e.setMolecularSequence( BasicSequence.createAaSequence( e.getAccession(), sq_buffer.toString() ) );
+        if ( sq_buffer.length() > 0 ) {
+            if ( is_aa ) {
+                e.setMolecularSequence( BasicSequence.createAaSequence( e.getAccession(), sq_buffer.toString() ) );
+            }
+            else {
+                e.setMolecularSequence( BasicSequence.createDnaSequence( e.getAccession(), sq_buffer.toString() ) );
+            }
         }
         return e;
     }
-
-    @Override
-    public SortedSet<Annotation> getAnnotations() {
-        return null;
-    }
-
-    @Override
-    public String getMap() {
-        return null;
-    }
-
-    @Override
-    public String getChromosome() {
-        return null;
-    }
-
-    @Override
-    public MolecularSequence getMolecularSequence() {
-        return _mol_seq;
-    }
 }