JAL-1705 refactoring etc for fetching Ensembl --> Uniprot

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

diff --git a/src/jalview/ext/ensembl/EnsemblCds.java b/src/jalview/ext/ensembl/EnsemblCds.java
index 1f63e05..2086eba 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblCds.java
+++ b/src/jalview/ext/ensembl/EnsemblCds.java
@@ -1,15 +1,51 @@
 package jalview.ext.ensembl;
 
-import jalview.ext.ensembl.SeqFetcher.EnsemblSeqType;
+import jalview.datamodel.SequenceFeature;
+import jalview.datamodel.SequenceI;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
 
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * A client for direct fetching of CDS sequences from Ensembl (i.e. that part of
+ * the genomic sequence that is translated to protein)
+ * 
+ * TODO: not currently used as CDS sequences are computed from CDS features on
+ * transcripts - delete this class?
+ * 
+ * @author gmcarstairs
+ *
+ */
 public class EnsemblCds extends EnsemblSeqProxy
 {
+  /*
+   * fetch cds features on genomic sequence (to identify the CDS regions)
+   * and exon and variation features (to retain for display)
+   */
+  private static final EnsemblFeatureType[] FEATURES_TO_FETCH = {
+      EnsemblFeatureType.cds, EnsemblFeatureType.exon,
+      EnsemblFeatureType.variation };
 
-  public EnsemblCds() throws Exception
+  /**
+   * Default constructor (to use rest.ensembl.org)
+   */
+  public EnsemblCds()
   {
     super();
   }
 
+  /**
+   * Constructor given the target domain to fetch data from
+   * 
+   * @param d
+   */
+  public EnsemblCds(String d)
+  {
+    super(d);
+  }
+
   @Override
   public String getDbName()
   {
@@ -22,4 +58,61 @@ public class EnsemblCds extends EnsemblSeqProxy
     return EnsemblSeqType.CDS;
   }
 
+  @Override
+  protected EnsemblFeatureType[] getFeaturesToFetch()
+  {
+    return FEATURES_TO_FETCH;
+  }
+
+  /**
+   * Answers true unless the feature type is 'CDS' (or a sub-type of CDS in the
+   * Sequence Ontology). CDS features are only retrieved in order to identify
+   * the cds sequence range, and are redundant information on the cds sequence
+   * itself.
+   */
+  @Override
+  protected boolean retainFeature(SequenceFeature sf, String accessionId)
+  {
+    if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+            SequenceOntologyI.CDS))
+    {
+      return false;
+    }
+    return featureMayBelong(sf, accessionId);
+  }
+
+  /**
+   * Answers true if the sequence feature type is 'CDS' (or a subtype of CDS in
+   * the Sequence Ontology), and the Parent of the feature is the transcript we
+   * are retrieving
+   */
+  @Override
+  protected boolean identifiesSequence(SequenceFeature sf, String accId)
+  {
+    if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+            SequenceOntologyI.CDS))
+    {
+      String parentFeature = (String) sf.getValue(PARENT);
+      if (("transcript:" + accId).equals(parentFeature))
+      {
+        return true;
+      }
+    }
+    return false;
+  }
+
+  /**
+   * Overrides this method to trivially return a range which is the whole of the
+   * nucleotide sequence. This is both faster than scanning for CDS features,
+   * and also means we don't need to keep CDS features on CDS sequence (where
+   * they are redundant information).
+   */
+  protected List<int[]> getCdsRanges(SequenceI dnaSeq)
+  {
+    int len = dnaSeq.getLength();
+    List<int[]> ranges = new ArrayList<int[]>();
+    ranges.add(new int[] { 1, len });
+    return ranges;
+  }
+
 }