JAL-2743 fall back to parsing description if chromosome lookup fails;

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

diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java
index f46cb1b..cdcfa96 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -23,6 +23,8 @@ package jalview.ext.ensembl;
 import jalview.api.FeatureColourI;
 import jalview.api.FeatureSettingsModelI;
 import jalview.datamodel.AlignmentI;
+import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.GeneLociI;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
@@ -144,8 +146,10 @@ public class EnsemblGene extends EnsemblSeqProxy
       {
         continue;
       }
+      
       if (geneAlignment.getHeight() == 1)
       {
+        findGeneLoci(geneAlignment.getSequenceAt(0), geneId);
         getTranscripts(geneAlignment, geneId);
       }
       if (al == null)
@@ -161,6 +165,67 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
+   * Calls the /lookup/id REST service, parses the response for gene
+   * coordinates, and if successful, adds these to the sequence. If this fails,
+   * fall back on trying to parse the sequence description in case it is in
+   * Ensembl-gene format e.g. chromosome:GRCh38:17:45051610:45109016:1.
+   * 
+   * @param seq
+   * @param geneId
+   */
+  void findGeneLoci(SequenceI seq, String geneId)
+  {
+    GeneLociI geneLoci = new EnsemblLookup(getDomain()).getGeneLoci(geneId);
+    if (geneLoci != null)
+    {
+      seq.setGeneLoci(geneLoci.getSpeciesId(), geneLoci.getAssemblyId(),
+              geneLoci.getChromosomeId(), geneLoci.getMap());
+    }
+    else
+    {
+      parseChromosomeLocations(seq);
+    }
+  }
+
+  /**
+   * Parses and saves fields of an Ensembl-style description e.g.
+   * chromosome:GRCh38:17:45051610:45109016:1
+   * 
+   * @param seq
+   */
+  boolean parseChromosomeLocations(SequenceI seq)
+  {
+    String description = seq.getDescription();
+    if (description == null)
+    {
+      return false;
+    }
+    String[] tokens = description.split(":");
+    if (tokens.length == 6 && tokens[0].startsWith(DBRefEntry.CHROMOSOME))
+    {
+      String ref = tokens[1];
+      String chrom = tokens[2];
+      try
+      {
+        int chStart = Integer.parseInt(tokens[3]);
+        int chEnd = Integer.parseInt(tokens[4]);
+        boolean forwardStrand = "1".equals(tokens[5]);
+        String species = ""; // not known here
+        int[] from = new int[] { seq.getStart(), seq.getEnd() };
+        int[] to = new int[] { forwardStrand ? chStart : chEnd,
+            forwardStrand ? chEnd : chStart };
+        MapList map = new MapList(from, to, 1, 1);
+        seq.setGeneLoci(species, ref, chrom, map);
+        return true;
+      } catch (NumberFormatException e)
+      {
+        System.err.println("Bad integers in description " + description);
+      }
+    }
+    return false;
+  }
+
+  /**
    * Converts a query, which may contain one or more gene, transcript, or
    * external (to Ensembl) identifiers, into a non-redundant list of gene
    * identifiers.
@@ -354,6 +419,8 @@ public class EnsemblGene extends EnsemblSeqProxy
     cdna.transferFeatures(gene.getFeatures().getPositionalFeatures(),
             transcript.getDatasetSequence(), mapping, parentId);
 
+    mapTranscriptToChromosome(transcript, gene, mapping);
+
     /*
      * fetch and save cross-references
      */
@@ -368,6 +435,42 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
+   * If the gene has a mapping to chromosome coordinates, derive the transcript
+   * chromosome regions and save on the transcript sequence
+   * 
+   * @param transcript
+   * @param gene
+   * @param mapping
+   *          the mapping from gene to transcript positions
+   */
+  protected void mapTranscriptToChromosome(SequenceI transcript,
+          SequenceI gene, MapList mapping)
+  {
+    GeneLociI loci = gene.getGeneLoci();
+    if (loci == null)
+    {
+      return;
+    }
+
+    MapList geneMapping = loci.getMap();
+
+    List<int[]> exons = mapping.getFromRanges();
+    List<int[]> transcriptLoci = new ArrayList<>();
+
+    for (int[] exon : exons)
+    {
+      transcriptLoci.add(geneMapping.locateInTo(exon[0], exon[1]));
+    }
+
+    List<int[]> transcriptRange = Arrays.asList(new int[] {
+        transcript.getStart(), transcript.getEnd() });
+    MapList mapList = new MapList(transcriptRange, transcriptLoci, 1, 1);
+
+    transcript.setGeneLoci(loci.getSpeciesId(), loci.getAssemblyId(),
+            loci.getChromosomeId(), mapList);
+  }
+
+  /**
    * Returns the 'transcript_id' property of the sequence feature (or null)
    * 
    * @param feature
@@ -381,6 +484,12 @@ public class EnsemblGene extends EnsemblSeqProxy
   /**
    * Returns a list of the transcript features on the sequence whose Parent is
    * the gene for the accession id.
+   * <p>
+   * Transcript features are those of type "transcript", or any of its sub-types
+   * in the Sequence Ontology e.g. "mRNA", "processed_transcript". We also
+   * include "NMD_transcript_variant", because this type behaves like a
+   * transcript identifier in Ensembl, although strictly speaking it is not in
+   * the SO.
    * 
    * @param accId
    * @param geneSequence
@@ -392,19 +501,18 @@ public class EnsemblGene extends EnsemblSeqProxy
     List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
 
     String parentIdentifier = GENE_PREFIX + accId;
-    // todo optimise here by transcript type!
+
     List<SequenceFeature> sfs = geneSequence.getFeatures()
-            .getPositionalFeatures();
+            .getFeaturesByOntology(SequenceOntologyI.TRANSCRIPT);
+    sfs.addAll(geneSequence.getFeatures().getPositionalFeatures(
+            SequenceOntologyI.NMD_TRANSCRIPT_VARIANT));
 
     for (SequenceFeature sf : sfs)
     {
-      if (isTranscript(sf.getType()))
+      String parent = (String) sf.getValue(PARENT);
+      if (parentIdentifier.equals(parent))
       {
-        String parent = (String) sf.getValue(PARENT);
-        if (parentIdentifier.equals(parent))
-        {
-          transcriptFeatures.add(sf);
-        }
+        transcriptFeatures.add(sf);
       }
     }