X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblGene.java;h=afff4c2630609be55291630e560b3c23e0bdbf05;hb=a57976ba40e1abe6d7c1940386e1a25419ef9c9d;hp=2d4d61abfa27a14aa4f5cb44946c3242b85e98dd;hpb=19fe93b91d4c50834b1a7101da2b1c240c55ae84;p=jalview.git

diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java
index 2d4d61a..afff4c2 100644
--- 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;
@@ -110,7 +112,8 @@ public class EnsemblGene extends EnsemblSeqProxy
    * <li>resolves an external identifier by looking up xref-ed gene ids</li>
    * <li>fetches the gene sequence</li>
    * <li>fetches features on the sequence</li>
-   * <li>identifies "transcript" features whose Parent is the requested gene</li>
+   * <li>identifies "transcript" features whose Parent is the requested
+   * gene</li>
    * <li>fetches the transcript sequence for each transcript</li>
    * <li>makes a mapping from the gene to each transcript</li>
    * <li>copies features from gene to transcript sequences</li>
@@ -143,6 +146,9 @@ public class EnsemblGene extends EnsemblSeqProxy
       {
         continue;
       }
+      
+      parseChromosomeLocations(geneAlignment);
+      
       if (geneAlignment.getHeight() == 1)
       {
         getTranscripts(geneAlignment, geneId);
@@ -160,8 +166,48 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Converts a query, which may contain one or more gene or transcript
-   * identifiers, into a non-redundant list of gene identifiers.
+   * Parses and saves fields of an Ensembl-style description e.g.
+   * chromosome:GRCh38:17:45051610:45109016:1
+   * 
+   * @param alignment
+   */
+  private void parseChromosomeLocations(AlignmentI alignment)
+  {
+    for (SequenceI seq : alignment.getSequences())
+    {
+      String description = seq.getDescription();
+      if (description == null)
+      {
+        continue;
+      }
+      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 = ""; // dunno yet!
+          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);
+        } catch (NumberFormatException e)
+        {
+          System.err.println("Bad integers in description " + description);
+        }
+      }
+    }
+  }
+
+  /**
+   * Converts a query, which may contain one or more gene, transcript, or
+   * external (to Ensembl) identifiers, into a non-redundant list of gene
+   * identifiers.
    * 
    * @param accessions
    * @return
@@ -172,39 +218,30 @@ public class EnsemblGene extends EnsemblSeqProxy
 
     for (String acc : accessions.split(getAccessionSeparator()))
     {
-      if (isGeneIdentifier(acc))
-      {
-        if (!geneIds.contains(acc))
-        {
-          geneIds.add(acc);
-        }
-      }
-
       /*
-       * if given a transcript id, look up its gene parent
+       * First try lookup as an Ensembl (gene or transcript) identifier
        */
-      else if (isTranscriptIdentifier(acc))
+      String geneId = new EnsemblLookup(getDomain()).getGeneId(acc);
+      if (geneId != null)
       {
-        String geneId = new EnsemblLookup(getDomain()).getParent(acc);
-        if (geneId != null && !geneIds.contains(geneId))
+        if (!geneIds.contains(geneId))
         {
           geneIds.add(geneId);
         }
       }
-
-      /*
-       * if given a gene or other external name, lookup and fetch 
-       * the corresponding gene for all model organisms 
-       */
       else
       {
+        /*
+         * if given a gene or other external name, lookup and fetch 
+         * the corresponding gene for all model organisms 
+         */
         List<String> ids = new EnsemblSymbol(getDomain(), getDbSource(),
-                getDbVersion()).getIds(acc);
-        for (String geneId : ids)
+                getDbVersion()).getGeneIds(acc);
+        for (String id : ids)
         {
-          if (!geneIds.contains(geneId))
+          if (!geneIds.contains(id))
           {
-            geneIds.add(geneId);
+            geneIds.add(id);
           }
         }
       }
@@ -213,30 +250,6 @@ public class EnsemblGene extends EnsemblSeqProxy
   }
 
   /**
-   * Attempts to get Ensembl stable identifiers for model organisms for a gene
-   * name by calling the xrefs symbol REST service to resolve the gene name.
-   * 
-   * @param query
-   * @return
-   */
-  protected String getGeneIdentifiersForName(String query)
-  {
-    List<String> ids = new EnsemblSymbol(getDomain(), getDbSource(),
-            getDbVersion()).getIds(query);
-    if (ids != null)
-    {
-      for (String id : ids)
-      {
-        if (isGeneIdentifier(id))
-        {
-          return id;
-        }
-      }
-    }
-    return null;
-  }
-
-  /**
    * Constructs all transcripts for the gene, as identified by "transcript"
    * features whose Parent is the requested gene. The coding transcript
    * sequences (i.e. with introns omitted) are added to the alignment.
@@ -298,8 +311,8 @@ public class EnsemblGene extends EnsemblSeqProxy
    *          the parent gene sequence, with features
    * @return
    */
-  SequenceI makeTranscript(SequenceFeature transcriptFeature,
-          AlignmentI al, SequenceI gene)
+  SequenceI makeTranscript(SequenceFeature transcriptFeature, AlignmentI al,
+          SequenceI gene)
   {
     String accId = getTranscriptId(transcriptFeature);
     if (accId == null)
@@ -348,7 +361,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       mappedFrom.add(new int[] { sf.getBegin(), sf.getEnd() });
     }
 
-    Sequence transcript = new Sequence(accId, seqChars, 1, transcriptLength);
+    Sequence transcript = new Sequence(accId, seqChars, 1,
+            transcriptLength);
 
     /*
      * Ensembl has gene name as transcript Name
@@ -384,6 +398,8 @@ public class EnsemblGene extends EnsemblSeqProxy
     cdna.transferFeatures(gene.getFeatures().getPositionalFeatures(),
             transcript.getDatasetSequence(), mapping, parentId);
 
+    mapTranscriptToChromosome(transcript, gene, mapping);
+
     /*
      * fetch and save cross-references
      */
@@ -398,6 +414,42 @@ 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
@@ -411,6 +463,12 @@ 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
@@ -422,19 +480,18 @@ 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);
       }
     }
 
@@ -551,8 +608,8 @@ public class EnsemblGene extends EnsemblSeqProxy
       @Override
       public boolean isFeatureDisplayed(String type)
       {
-        return (so.isA(type, SequenceOntologyI.EXON) || so.isA(type,
-                SequenceOntologyI.SEQUENCE_VARIANT));
+        return (so.isA(type, SequenceOntologyI.EXON)
+                || so.isA(type, SequenceOntologyI.SEQUENCE_VARIANT));
       }
 
       @Override