Jalview-JS/JAL-3253-applet also comments relating to JAL-3268

[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 7e6f653..63a6a6c 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblGene.java
+++ b/src/jalview/ext/ensembl/EnsemblGene.java
@@ -34,6 +34,7 @@ import jalview.io.gff.SequenceOntologyI;
 import jalview.schemes.FeatureColour;
 import jalview.schemes.FeatureSettingsAdapter;
 import jalview.util.MapList;
+import jalview.util.Platform;
 
 import java.awt.Color;
 import java.io.UnsupportedEncodingException;
@@ -51,13 +52,11 @@ import com.stevesoft.pat.Regex;
  */
 public class EnsemblGene extends EnsemblSeqProxy
 {
-  private static final String GENE_PREFIX = "gene:";
-
   /*
    * accepts anything as we will attempt lookup of gene or 
    * transcript id or gene name
    */
-  private static final Regex ACCESSION_REGEX = new Regex(".*");
+  private static Regex ACCESSION_REGEX;
 
   private static final EnsemblFeatureType[] FEATURES_TO_FETCH = {
       EnsemblFeatureType.gene, EnsemblFeatureType.transcript,
@@ -140,7 +139,6 @@ public class EnsemblGene extends EnsemblSeqProxy
      * convert to a non-duplicated list of gene identifiers
      */
     List<String> geneIds = getGeneIds(query);
-
     AlignmentI al = null;
     for (String geneId : geneIds)
     {
@@ -157,9 +155,7 @@ public class EnsemblGene extends EnsemblSeqProxy
       {
         // 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)
@@ -368,7 +364,7 @@ public class EnsemblGene extends EnsemblSeqProxy
      * look for exon features of the transcript, failing that for CDS
      * (for example ENSG00000124610 has 1 CDS but no exon features)
      */
-    String parentId = "transcript:" + accId;
+    String parentId = accId;
     List<SequenceFeature> splices = findFeatures(gene,
             SequenceOntologyI.EXON, parentId);
     if (splices.isEmpty())
@@ -399,7 +395,7 @@ public class EnsemblGene extends EnsemblSeqProxy
      * Ensembl has gene name as transcript Name
      * EnsemblGenomes doesn't, but has a url-encoded description field
      */
-    String description = (String) transcriptFeature.getValue(NAME);
+    String description = transcriptFeature.getDescription();
     if (description == null)
     {
       description = (String) transcriptFeature.getValue(DESCRIPTION);
@@ -488,7 +484,7 @@ public class EnsemblGene extends EnsemblSeqProxy
    */
   protected String getTranscriptId(SequenceFeature feature)
   {
-    return (String) feature.getValue("transcript_id");
+    return (String) feature.getValue(JSON_ID);
   }
 
   /**
@@ -510,7 +506,7 @@ public class EnsemblGene extends EnsemblSeqProxy
   {
     List<SequenceFeature> transcriptFeatures = new ArrayList<>();
 
-    String parentIdentifier = GENE_PREFIX + accId;
+    String parentIdentifier = accId;
 
     List<SequenceFeature> sfs = geneSequence.getFeatures()
             .getFeaturesByOntology(SequenceOntologyI.TRANSCRIPT);
@@ -541,30 +537,34 @@ public class EnsemblGene extends EnsemblSeqProxy
   @Override
   public String getTestQuery()
   {
-    return "ENSG00000157764"; // BRAF, 5 transcripts, reverse strand
+    return Platform.isJS() ? "ENSG00000123569" : "ENSG00000157764";
+    // ENSG00000123569 // H2BFWT histone, 2 transcripts, reverse strand
+    // ENSG00000157764 // BRAF, 5 transcripts, reverse strand
     // ENSG00000090266 // NDUFB2, 15 transcripts, forward strand
     // ENSG00000101812 // H2BFM histone, 3 transcripts, forward strand
-    // ENSG00000123569 // H2BFWT histone, 2 transcripts, reverse strand
   }
 
   /**
-   * 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).equalsIgnoreCase(id))
+      String id = (String) sf.getValue(JSON_ID);
+      if (accId.equalsIgnoreCase(id))
       {
-        return true;
+        result.add(sf);
       }
     }
-    return false;
+    return result;
   }
 
   /**
@@ -577,7 +577,7 @@ public class EnsemblGene extends EnsemblSeqProxy
   @Override
   protected boolean retainFeature(SequenceFeature sf, String accessionId)
   {
-    SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+    SequenceOntologyI so = SequenceOntologyFactory.getSequenceOntology();
     String type = sf.getType();
     if (so.isA(type, SequenceOntologyI.GENE))
     {
@@ -586,7 +586,7 @@ public class EnsemblGene extends EnsemblSeqProxy
     if (isTranscript(type))
     {
       String parent = (String) sf.getValue(PARENT);
-      if (!(GENE_PREFIX + accessionId).equalsIgnoreCase(parent))
+      if (!accessionId.equalsIgnoreCase(parent))
       {
         return false;
       }
@@ -595,17 +595,6 @@ 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
    */
@@ -617,6 +606,10 @@ public class EnsemblGene extends EnsemblSeqProxy
   @Override
   public Regex getAccessionValidator()
   {
+    if (ACCESSION_REGEX == null)
+    {
+      ACCESSION_REGEX = Platform.newRegex(".*", null);
+    }
     return ACCESSION_REGEX;
   }
 
@@ -635,7 +628,7 @@ public class EnsemblGene extends EnsemblSeqProxy
   {
     return new FeatureSettingsAdapter()
     {
-      SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+      SequenceOntologyI so = SequenceOntologyFactory.getSequenceOntology();
 
       @Override
       public boolean isFeatureDisplayed(String type)