Merge branch 'bug/JAL-3120restoreFeatureColour' into merge/JAL-3120

[jalview.git] / test / jalview / ext / ensembl / EnsemblGeneTest.java

diff --git a/test/jalview/ext/ensembl/EnsemblGeneTest.java b/test/jalview/ext/ensembl/EnsemblGeneTest.java
index 5920b89..8b1e840 100644 (file)
--- a/test/jalview/ext/ensembl/EnsemblGeneTest.java
+++ b/test/jalview/ext/ensembl/EnsemblGeneTest.java
@@ -25,6 +25,8 @@ import static org.testng.AssertJUnit.assertFalse;
 import static org.testng.AssertJUnit.assertTrue;
 
 import jalview.api.FeatureSettingsModelI;
+import jalview.bin.Cache;
+import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceDummy;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
@@ -53,6 +55,7 @@ public class EnsemblGeneTest
   @BeforeClass(alwaysRun = true)
   public void setUp()
   {
+    Cache.loadProperties("test/jalview/io/testProps.jvprops");
     SequenceOntologyFactory.setInstance(new SequenceOntologyLite());
   }
 
@@ -75,18 +78,10 @@ public class EnsemblGeneTest
     genomic.setEnd(50000);
     String geneId = "ABC123";
 
-    // gene at (start+20000) length 501
-    // should be ignored - the first 'gene' found defines the whole range
-    // (note features are found in position order, not addition order)
-    SequenceFeature sf = new SequenceFeature("gene", "", 20000, 20500, 0f,
-            null);
-    sf.setValue("ID", "gene:" + geneId);
-    sf.setStrand("+");
-    genomic.addSequenceFeature(sf);
-
     // gene at (start + 10500) length 101
-    sf = new SequenceFeature("gene", "", 10500, 10600, 0f, null);
-    sf.setValue("ID", "gene:" + geneId);
+    SequenceFeature sf = new SequenceFeature("gene", "", 10500, 10600, 0f,
+            null);
+    sf.setValue("id", geneId);
     sf.setStrand("+");
     genomic.addSequenceFeature(sf);
 
@@ -115,18 +110,10 @@ public class EnsemblGeneTest
     genomic.setEnd(50000);
     String geneId = "ABC123";
 
-    // gene at (start+20000) length 501
-    // should be ignored - the first 'gene' found defines the whole range
-    // (real data would only have one such feature)
-    SequenceFeature sf = new SequenceFeature("ncRNA_gene", "", 20000,
-            20500, 0f, null);
-    sf.setValue("ID", "gene:" + geneId);
-    sf.setStrand("-");
-    genomic.addSequenceFeature(sf);
-
     // gene at (start + 10500) length 101
-    sf = new SequenceFeature("gene", "", 10500, 10600, 0f, null);
-    sf.setValue("ID", "gene:" + geneId);
+    SequenceFeature sf = new SequenceFeature("gene", "", 10500, 10600, 0f,
+            null);
+    sf.setValue("id", geneId);
     sf.setStrand("+");
     genomic.addSequenceFeature(sf);
 
@@ -159,29 +146,30 @@ public class EnsemblGeneTest
     // transcript feature
     SequenceFeature sf1 = new SequenceFeature("transcript", "", 20000,
             20500, 0f, null);
-    sf1.setValue("Parent", "gene:" + geneId);
-    sf1.setValue("transcript_id", "transcript1");
+    sf1.setValue("Parent", geneId);
+    sf1.setValue("id", "transcript1");
     genomic.addSequenceFeature(sf1);
 
     // transcript sub-type feature
     SequenceFeature sf2 = new SequenceFeature("snRNA", "", 21000, 21500,
             0f, null);
-    sf2.setValue("Parent", "gene:" + geneId);
-    sf2.setValue("transcript_id", "transcript2");
+    sf2.setValue("Parent", geneId);
+    sf2.setValue("id", "transcript2");
     genomic.addSequenceFeature(sf2);
 
     // NMD_transcript_variant treated like transcript in Ensembl
     SequenceFeature sf3 = new SequenceFeature("NMD_transcript_variant", "",
             22000, 22500, 0f, null);
-    sf3.setValue("Parent", "gene:" + geneId);
-    sf3.setValue("transcript_id", "transcript3");
+    // id matching should not be case-sensitive
+    sf3.setValue("Parent", geneId.toLowerCase());
+    sf3.setValue("id", "transcript3");
     genomic.addSequenceFeature(sf3);
 
     // transcript for a different gene - ignored
     SequenceFeature sf4 = new SequenceFeature("snRNA", "", 23000, 23500,
             0f, null);
-    sf4.setValue("Parent", "gene:XYZ");
-    sf4.setValue("transcript_id", "transcript4");
+    sf4.setValue("Parent", "XYZ");
+    sf4.setValue("id", "transcript4");
     genomic.addSequenceFeature(sf4);
 
     EnsemblGene testee = new EnsemblGene();
@@ -208,24 +196,24 @@ public class EnsemblGeneTest
     EnsemblGene testee = new EnsemblGene();
     SequenceFeature sf = new SequenceFeature("gene", "", 20000, 20500, 0f,
             null);
-    sf.setValue("ID", "gene:" + geneId);
+    sf.setValue("id", geneId);
     assertFalse(testee.retainFeature(sf, geneId));
 
     sf = new SequenceFeature("transcript", "", 20000, 20500, 0f, null);
-    sf.setValue("Parent", "gene:" + geneId);
+    sf.setValue("Parent", geneId);
     assertTrue(testee.retainFeature(sf, geneId));
 
     sf = new SequenceFeature("mature_transcript", "", 20000, 20500, 0f,
             null);
-    sf.setValue("Parent", "gene:" + geneId);
+    sf.setValue("Parent", geneId);
     assertTrue(testee.retainFeature(sf, geneId));
 
     sf = new SequenceFeature("NMD_transcript_variant", "", 20000, 20500,
             0f, null);
-    sf.setValue("Parent", "gene:" + geneId);
+    sf.setValue("Parent", geneId);
     assertTrue(testee.retainFeature(sf, geneId));
 
-    sf.setValue("Parent", "gene:XYZ");
+    sf.setValue("Parent", "ßXYZ");
     assertFalse(testee.retainFeature(sf, geneId));
 
     sf = new SequenceFeature("anything", "", 20000, 20500, 0f, null);
@@ -237,37 +225,48 @@ public class EnsemblGeneTest
    * accession id as ID
    */
   @Test(groups = "Functional")
-  public void testIdentifiesSequence()
+  public void testGetIdentifyingFeatures()
   {
     String accId = "ABC123";
-    EnsemblGene testee = new EnsemblGene();
+    SequenceI seq = new Sequence(accId, "HIBEES");
 
     // gene with no ID not valid
-    SequenceFeature sf = new SequenceFeature("gene", "", 1, 2, 0f, null);
-    assertFalse(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf1 = new SequenceFeature("gene", "", 1, 2, 0f, null);
+    seq.addSequenceFeature(sf1);
 
     // gene with wrong ID not valid
-    sf.setValue("ID", "gene:XYZ");
-    assertFalse(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf2 = new SequenceFeature("gene", "a", 1, 2, 0f, null);
+    sf2.setValue("id", "XYZ");
+    seq.addSequenceFeature(sf2);
 
     // gene with right ID is valid
-    sf.setValue("ID", "gene:" + accId);
-    assertTrue(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf3 = new SequenceFeature("gene", "b", 1, 2, 0f, null);
+    sf3.setValue("id", accId);
+    seq.addSequenceFeature(sf3);
 
     // gene sub-type with right ID is valid
-    sf = new SequenceFeature("snRNA_gene", "", 1, 2, 0f, null);
-    sf.setValue("ID", "gene:" + accId);
-    assertTrue(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf4 = new SequenceFeature("snRNA_gene", "", 1, 2, 0f, null);
+    sf4.setValue("id", accId);
+    seq.addSequenceFeature(sf4);
 
     // transcript not valid:
-    sf = new SequenceFeature("transcript", "", 1, 2, 0f, null);
-    sf.setValue("ID", "gene:" + accId);
-    assertFalse(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf5 = new SequenceFeature("transcript", "", 1, 2, 0f, null);
+    sf5.setValue("id", accId);
+    seq.addSequenceFeature(sf5);
 
     // exon not valid:
-    sf = new SequenceFeature("exon", "", 1, 2, 0f, null);
-    sf.setValue("ID", "gene:" + accId);
-    assertFalse(testee.identifiesSequence(sf, accId));
+    SequenceFeature sf6 = new SequenceFeature("exon", "", 1, 2, 0f, null);
+    sf6.setValue("id", accId);
+    seq.addSequenceFeature(sf6);
+    
+    List<SequenceFeature> sfs = new EnsemblGene()
+            .getIdentifyingFeatures(seq, accId);
+    assertFalse(sfs.contains(sf1));
+    assertFalse(sfs.contains(sf2));
+    assertTrue(sfs.contains(sf3));
+    assertTrue(sfs.contains(sf4));
+    assertFalse(sfs.contains(sf5));
+    assertFalse(sfs.contains(sf6));
   }
 
   /**