X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblGene.java;h=6923b774c21ddf844643b51ecc8f6fd35d677894;hb=refs%2Fheads%2Ffeature%2FJAL-2738GeneLocus;hp=919134ccb3e267b974c6135d7648666409f518eb;hpb=17ff1f476e009b3a3c7e892e416edc2a4af8a2bc;p=jalview.git

diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java
index 919134c..6923b77 100644
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -23,6 +23,7 @@ package jalview.ext.ensembl;
 import jalview.api.FeatureColourI;
 import jalview.api.FeatureSettingsModelI;
 import jalview.datamodel.AlignmentI;
+import jalview.datamodel.GeneLociI;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
@@ -62,6 +63,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       EnsemblFeatureType.exon, EnsemblFeatureType.cds,
       EnsemblFeatureType.variation };
 
+  private static final String CHROMOSOME = "chromosome";
+
   /**
    * Default constructor (to use rest.ensembl.org)
    */
@@ -150,8 +153,14 @@ public class EnsemblGene extends EnsemblSeqProxy
       {
         continue;
       }
+      
       if (geneAlignment.getHeight() == 1)
       {
+        // ensure id has 'correct' case for the Ensembl identifier
+        geneId = geneAlignment.getSequenceAt(0).getName();
+
+        findGeneLoci(geneAlignment.getSequenceAt(0), geneId);
+
         getTranscripts(geneAlignment, geneId);
       }
       if (al == null)
@@ -167,6 +176,67 @@ 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.getMapping());
+    }
+    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(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.
@@ -360,6 +430,8 @@ public class EnsemblGene extends EnsemblSeqProxy
     cdna.transferFeatures(gene.getFeatures().getPositionalFeatures(),
             transcript.getDatasetSequence(), mapping, parentId);
 
+    mapTranscriptToChromosome(transcript, gene, mapping);
+
     /*
      * fetch and save cross-references
      */
@@ -374,6 +446,42 @@ 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.getMapping();
+
+    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
@@ -413,7 +521,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     for (SequenceFeature sf : sfs)
     {
       String parent = (String) sf.getValue(PARENT);
-      if (parentIdentifier.equals(parent))
+      if (parentIdentifier.equalsIgnoreCase(parent))
       {
         transcriptFeatures.add(sf);
       }
@@ -441,23 +549,27 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Answers true for a feature of type 'gene' (or a sub-type of gene in the
-   * Sequence Ontology), whose ID is the accession we are retrieving
+   * Answers a list of sequence features (if any) whose type is 'gene' (or a
+   * subtype of gene in the Sequence Ontology), and whose ID is the accession we
+   * are retrieving
    */
   @Override
-  protected boolean identifiesSequence(SequenceFeature sf, String accId)
+  protected List<SequenceFeature> getIdentifyingFeatures(SequenceI seq,
+          String accId)
   {
-    if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
-            SequenceOntologyI.GENE))
+    List<SequenceFeature> result = new ArrayList<>();
+    List<SequenceFeature> sfs = seq.getFeatures()
+            .getFeaturesByOntology(SequenceOntologyI.GENE);
+    for (SequenceFeature sf : sfs)
     {
       // NB features as gff use 'ID'; rest services return as 'id'
       String id = (String) sf.getValue("ID");
-      if ((GENE_PREFIX + accId).equals(id))
+      if ((GENE_PREFIX + accId).equalsIgnoreCase(id))
       {
-        return true;
+        result.add(sf);
       }
     }
-    return false;
+    return result;
   }
 
   /**
@@ -479,7 +591,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (isTranscript(type))
     {
       String parent = (String) sf.getValue(PARENT);
-      if (!(GENE_PREFIX + accessionId).equals(parent))
+      if (!(GENE_PREFIX + accessionId).equalsIgnoreCase(parent))
       {
         return false;
       }
@@ -488,17 +600,6 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Answers false. This allows an optimisation - a single 'gene' feature is all
-   * that is needed to identify the positions of the gene on the genomic
-   * sequence.
-   */
-  @Override
-  protected boolean isSpliceable()
-  {
-    return false;
-  }
-
-  /**
    * Override to do nothing as Ensembl doesn't return a protein sequence for a
    * gene identifier
    */