JAL-1705 various refactoring towards Uniprot-to-Ensembl fetching

[jalview.git] / src / jalview / ext / ensembl / EnsemblCdna.java

diff --git a/src/jalview/ext/ensembl/EnsemblCdna.java b/src/jalview/ext/ensembl/EnsemblCdna.java
index b8c9c3f..f60125b 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblCdna.java
+++ b/src/jalview/ext/ensembl/EnsemblCdna.java
@@ -1,14 +1,21 @@
 package jalview.ext.ensembl;
 
 import jalview.datamodel.SequenceFeature;
-import jalview.io.gff.SequenceOntology;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
+
+import java.util.List;
 
 import com.stevesoft.pat.Regex;
 
 public class EnsemblCdna extends EnsemblSeqProxy
 {
+  // TODO modify to accept other species e.g. ENSMUSPnnn
+  private static final Regex ACCESSION_REGEX = new Regex(
+          "(ENST|ENSG|CCDS)[0-9.]{3,}$");
+  
   /*
-   * fetch exon features on genomic sequence (to identify the cdnaregions)
+   * fetch exon features on genomic sequence (to identify the cdna regions)
    * and cds and variation features (to retain)
    */
   private static final EnsemblFeatureType[] FEATURES_TO_FETCH = {
@@ -35,7 +42,7 @@ public class EnsemblCdna extends EnsemblSeqProxy
   @Override
   public Regex getAccessionValidator()
   {
-    return new Regex("((ENST|ENSG|CCDS)[0-9.]{3,})");
+    return ACCESSION_REGEX;
   }
 
   @Override
@@ -45,15 +52,17 @@ public class EnsemblCdna extends EnsemblSeqProxy
   }
 
   /**
-   * Answers true unless the feature type is 'exon' (or a sub-type of exon in
-   * the Sequence Ontology). Exon features are only retrieved in order to
-   * identify the exon sequence range, and are redundant information on the exon
-   * sequence itself.
+   * Answers true unless the feature type is 'transcript' (or a sub-type in the
+   * Sequence Ontology).
    */
   @Override
-  protected boolean retainFeature(String type)
+  protected boolean retainFeature(SequenceFeature sf, String accessionId)
   {
-    return !SequenceOntology.getInstance().isA(type, SequenceOntology.EXON);
+    if (isTranscript(sf.getType()))
+    {
+      return false;
+    }
+    return featureMayBelong(sf, accessionId);
   }
 
   /**
@@ -64,10 +73,10 @@ public class EnsemblCdna extends EnsemblSeqProxy
   @Override
   protected boolean identifiesSequence(SequenceFeature sf, String accId)
   {
-    if (SequenceOntology.getInstance().isA(sf.getType(),
-            SequenceOntology.EXON))
+    if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+            SequenceOntologyI.EXON))
     {
-      String parentFeature = (String) sf.getValue("Parent");
+      String parentFeature = (String) sf.getValue(PARENT);
       if (("transcript:" + accId).equals(parentFeature))
       {
         return true;
@@ -76,4 +85,12 @@ public class EnsemblCdna extends EnsemblSeqProxy
     return false;
   }
 
+  @Override
+  protected List<String> getCrossReferenceDatabases()
+  {
+    return super.getCrossReferenceDatabases();
+    // 30/01/16 also found Vega_transcript, OTTT, ENS_LRG_transcript, UCSC,
+    // HGNC_trans_name, RefSeq_mRNA, RefSeq_mRNA_predicted
+  }
+
 }