JAL-3076 refactor for more efficient scan of 'gene' features

[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 217742d..446b4f7 100644 (file)
--- a/test/jalview/ext/ensembl/EnsemblGeneTest.java
+++ b/test/jalview/ext/ensembl/EnsemblGeneTest.java
@@ -26,6 +26,7 @@ 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;
@@ -77,17 +78,9 @@ 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);
+    SequenceFeature sf = new SequenceFeature("gene", "", 10500, 10600, 0f,
+            null);
     sf.setValue("ID", "gene:" + geneId);
     sf.setStrand("+");
     genomic.addSequenceFeature(sf);
@@ -117,17 +110,9 @@ 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);
+    SequenceFeature sf = new SequenceFeature("gene", "", 10500, 10600, 0f,
+            null);
     sf.setValue("ID", "gene:" + geneId);
     sf.setStrand("+");
     genomic.addSequenceFeature(sf);
@@ -240,40 +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", "", 1, 2, 0f, null);
+    sf2.setValue("ID", "gene: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", "", 1, 2, 0f, null);
+    sf3.setValue("ID", "gene:" + 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));
-
-    // test is not case-sensitive
-    assertTrue(testee.identifiesSequence(sf, accId.toLowerCase()));
+    SequenceFeature sf4 = new SequenceFeature("snRNA_gene", "", 1, 2, 0f, null);
+    sf4.setValue("ID", "gene:" + 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", "gene:" + 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", "gene:" + 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));
   }
 
   /**