JAL-3143 fetch Ensembl(Genomes) features as JSON not GFF

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

diff --git a/src/jalview/ext/ensembl/EnsemblSeqProxy.java b/src/jalview/ext/ensembl/EnsemblSeqProxy.java
index f96f1d5..7a37c8a 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblSeqProxy.java
+++ b/src/jalview/ext/ensembl/EnsemblSeqProxy.java
@@ -62,8 +62,6 @@ import org.json.simple.parser.ParseException;
  */
 public abstract class EnsemblSeqProxy extends EnsemblRestClient
 {
-  protected static final String NAME = "Name";
-
   protected static final String DESCRIPTION = "description";
 
   /*
@@ -442,10 +440,10 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
   }
 
   /**
-   * Parses a JSON response into a list of sequences
+   * Parses a JSON response for a single sequence ID query
    * 
    * @param br
-   * @return
+   * @return a single jalview.datamodel.Sequence
    * @see http://rest.ensembl.org/documentation/info/sequence_id
    */
   protected List<SequenceI> parseSequenceJson(BufferedReader br)
@@ -581,8 +579,8 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
   protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
           String accId, int start)
   {
-    List<SequenceFeature> sfs = sourceSequence.getFeatures()
-            .getPositionalFeatures();
+    List<SequenceFeature> sfs = getIdentifyingFeatures(sourceSequence,
+            accId);
     if (sfs.isEmpty())
     {
       return null;
@@ -599,47 +597,31 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
 
     for (SequenceFeature sf : sfs)
     {
+      int strand = sf.getStrand();
+      strand = strand == 0 ? 1 : strand; // treat unknown as forward
+
+      if (directionSet && strand != direction)
+      {
+        // abort - mix of forward and backward
+        System.err
+                .println("Error: forward and backward strand for " + accId);
+        return null;
+      }
+      direction = strand;
+      directionSet = true;
+
       /*
-       * accept the target feature type or a specialisation of it
-       * (e.g. coding_exon for exon)
+       * add to CDS ranges, semi-sorted forwards/backwards
        */
-      if (identifiesSequence(sf, accId))
+      if (strand < 0)
       {
-        int strand = sf.getStrand();
-        strand = strand == 0 ? 1 : strand; // treat unknown as forward
-
-        if (directionSet && strand != direction)
-        {
-          // abort - mix of forward and backward
-          System.err.println(
-                  "Error: forward and backward strand for " + accId);
-          return null;
-        }
-        direction = strand;
-        directionSet = true;
-
-        /*
-         * add to CDS ranges, semi-sorted forwards/backwards
-         */
-        if (strand < 0)
-        {
-          regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
-        }
-        else
-        {
-          regions.add(new int[] { sf.getBegin(), sf.getEnd() });
-        }
-        mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
-
-        if (!isSpliceable())
-        {
-          /*
-           * 'gene' sequence is contiguous so we can stop as soon as its
-           * identifying feature has been found
-           */
-          break;
-        }
+        regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
+      }
+      else
+      {
+        regions.add(new int[] { sf.getBegin(), sf.getEnd() });
       }
+      mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
     }
 
     if (regions.isEmpty())
@@ -664,28 +646,6 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
   }
 
   /**
-   * Answers true if the sequence being retrieved may occupy discontiguous
-   * regions on the genomic sequence.
-   */
-  protected boolean isSpliceable()
-  {
-    return true;
-  }
-
-  /**
-   * Returns true if the sequence feature marks positions of the genomic
-   * sequence feature which are within the sequence being retrieved. For
-   * example, an 'exon' feature whose parent is the target transcript marks the
-   * cdna positions of the transcript.
-   * 
-   * @param sf
-   * @param accId
-   * @return
-   */
-  protected abstract boolean identifiesSequence(SequenceFeature sf,
-          String accId);
-
-  /**
    * Answers a list of sequence features that mark positions of the genomic
    * sequence feature which are within the sequence being retrieved. For
    * example, an 'exon' feature whose parent is the target transcript marks the
@@ -905,9 +865,8 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
   protected boolean featureMayBelong(SequenceFeature sf, String identifier)
   {
     String parent = (String) sf.getValue(PARENT);
-    // using contains to allow for prefix "gene:", "transcript:" etc
     if (parent != null
-            && !parent.toUpperCase().contains(identifier.toUpperCase()))
+            && !parent.equalsIgnoreCase(identifier))
     {
       // this genomic feature belongs to a different transcript
       return false;
@@ -915,6 +874,9 @@ public abstract class EnsemblSeqProxy extends EnsemblRestClient
     return true;
   }
 
+  /**
+   * Answers a short description of the sequence fetcher
+   */
   @Override
   public String getDescription()
   {