X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblCdna.java;h=e141db4b172d18876ba472a4eefc2baf788bdea2;hb=a7c519749ec05f55e11494c0d60df6e1af670a98;hp=0a9742536236ea97f7c6399a00b6a0fe40b22509;hpb=0b8abe58b934e03c34575b06d532a99f0ba70196;p=jalview.git

diff --git a/src/jalview/ext/ensembl/EnsemblCdna.java b/src/jalview/ext/ensembl/EnsemblCdna.java
index 0a97425..e141db4 100644
--- a/src/jalview/ext/ensembl/EnsemblCdna.java
+++ b/src/jalview/ext/ensembl/EnsemblCdna.java
@@ -1,25 +1,55 @@
 package jalview.ext.ensembl;
 
 import jalview.datamodel.SequenceFeature;
-import jalview.io.gff.SequenceOntology;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
 
 import com.stevesoft.pat.Regex;
 
+/**
+ * A client to fetch CDNA sequence from Ensembl (i.e. that part of the genomic
+ * sequence that is transcribed to RNA, but not necessarily translated to
+ * protein)
+ * 
+ * @author gmcarstairs
+ *
+ */
 public class EnsemblCdna extends EnsemblSeqProxy
 {
   /*
-   * fetch exon features on genomic sequence (to identify the cdnaregions)
+   * accepts ENST or ENSTG with 11 digits
+   * or ENSMUST or similar for other species
+   * or CCDSnnnnn.nn with at least 3 digits
+   */
+  private static final Regex ACCESSION_REGEX = new Regex(
+          "(ENS([A-Z]{3}|)[TG][0-9]{11}$)" + "|" + "(CCDS[0-9.]{3,}$)");
+  
+  /*
+   * 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 = {
       EnsemblFeatureType.exon, EnsemblFeatureType.cds,
       EnsemblFeatureType.variation };
 
+  /**
+   * Default constructor (to use rest.ensembl.org)
+   */
   public EnsemblCdna()
   {
     super();
   }
 
+  /**
+   * Constructor given the target domain to fetch data from
+   * 
+   * @param d
+   */
+  public EnsemblCdna(String d)
+  {
+    super(d);
+  }
+
   @Override
   public String getDbName()
   {
@@ -35,7 +65,7 @@ public class EnsemblCdna extends EnsemblSeqProxy
   @Override
   public Regex getAccessionValidator()
   {
-    return new Regex("((ENST|ENSG|CCDS)[0-9.]{3,})");
+    return ACCESSION_REGEX;
   }
 
   @Override
@@ -45,18 +75,13 @@ public class EnsemblCdna extends EnsemblSeqProxy
   }
 
   /**
-   * Answers true unless the feature type is 'transcript' or 'exon' (or a
-   * sub-type in the Sequence Ontology). These features are only retrieved in
-   * order to identify the exon sequence loci, 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(SequenceFeature sf, String accessionId)
   {
-    SequenceOntology so = SequenceOntology.getInstance();
-    String type = sf.getType();
-
-    if (isTranscript(type) || so.isA(type, SequenceOntology.EXON))
+    if (isTranscript(sf.getType()))
     {
       return false;
     }
@@ -71,8 +96,8 @@ 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);
       if (("transcript:" + accId).equals(parentFeature))