JAL-3019 correctly form toRange of 'traversed' MapList

[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 afff4c2..7e6f653 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -100,6 +100,12 @@ public class EnsemblGene extends EnsemblSeqProxy
     return EnsemblSeqType.GENOMIC;
   }
 
+  @Override
+  protected String getObjectType()
+  {
+    return OBJECT_TYPE_GENE;
+  }
+
   /**
    * Returns an alignment containing the gene(s) for the given gene or
    * transcript identifier, or external identifier (e.g. Uniprot id). If given a
@@ -147,10 +153,13 @@ public class EnsemblGene extends EnsemblSeqProxy
         continue;
       }
       
-      parseChromosomeLocations(geneAlignment);
-      
       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)
@@ -166,42 +175,64 @@ 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 alignment
+   * @param seq
    */
-  private void parseChromosomeLocations(AlignmentI alignment)
+  boolean parseChromosomeLocations(SequenceI seq)
   {
-    for (SequenceI seq : alignment.getSequences())
+    String description = seq.getDescription();
+    if (description == null)
     {
-      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
       {
-        continue;
-      }
-      String[] tokens = description.split(":");
-      if (tokens.length == 6 && tokens[0].startsWith(DBRefEntry.CHROMOSOME))
+        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)
       {
-        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);
-        }
+        System.err.println("Bad integers in description " + description);
       }
     }
+    return false;
   }
 
   /**
@@ -214,7 +245,7 @@ public class EnsemblGene extends EnsemblSeqProxy
    */
   List<String> getGeneIds(String accessions)
   {
-    List<String> geneIds = new ArrayList<String>();
+    List<String> geneIds = new ArrayList<>();
 
     for (String acc : accessions.split(getAccessionSeparator()))
     {
@@ -349,7 +380,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     int transcriptLength = 0;
     final char[] geneChars = gene.getSequence();
     int offset = gene.getStart(); // to convert to 0-based positions
-    List<int[]> mappedFrom = new ArrayList<int[]>();
+    List<int[]> mappedFrom = new ArrayList<>();
 
     for (SequenceFeature sf : splices)
     {
@@ -391,7 +422,7 @@ public class EnsemblGene extends EnsemblSeqProxy
      * transfer features to the new sequence; we use EnsemblCdna to do this,
      * to filter out unwanted features types (see method retainFeature)
      */
-    List<int[]> mapTo = new ArrayList<int[]>();
+    List<int[]> mapTo = new ArrayList<>();
     mapTo.add(new int[] { 1, transcriptLength });
     MapList mapping = new MapList(mappedFrom, mapTo, 1, 1);
     EnsemblCdna cdna = new EnsemblCdna(getDomain());
@@ -477,7 +508,7 @@ public class EnsemblGene extends EnsemblSeqProxy
   protected List<SequenceFeature> getTranscriptFeatures(String accId,
           SequenceI geneSequence)
   {
-    List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
+    List<SequenceFeature> transcriptFeatures = new ArrayList<>();
 
     String parentIdentifier = GENE_PREFIX + accId;
 
@@ -489,7 +520,7 @@ 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);
       }
@@ -526,8 +557,9 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
             SequenceOntologyI.GENE))
     {
-      String id = (String) sf.getValue(ID);
-      if ((GENE_PREFIX + accId).equals(id))
+      // NB features as gff use 'ID'; rest services return as 'id'
+      String id = (String) sf.getValue("ID");
+      if ((GENE_PREFIX + accId).equalsIgnoreCase(id))
       {
         return true;
       }
@@ -554,7 +586,7 @@ 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;
       }