JAL-2110 ensuring mapped dbrefs between protein and cds

[jalview.git] / test / jalview / analysis / AlignmentUtilsTests.java

diff --git a/test/jalview/analysis/AlignmentUtilsTests.java b/test/jalview/analysis/AlignmentUtilsTests.java
index 9394385..727cf22 100644 (file)
--- a/test/jalview/analysis/AlignmentUtilsTests.java
+++ b/test/jalview/analysis/AlignmentUtilsTests.java
@@ -22,6 +22,7 @@ package jalview.analysis;
 
 import static org.testng.AssertJUnit.assertEquals;
 import static org.testng.AssertJUnit.assertFalse;
+import static org.testng.AssertJUnit.assertNotNull;
 import static org.testng.AssertJUnit.assertNull;
 import static org.testng.AssertJUnit.assertSame;
 import static org.testng.AssertJUnit.assertTrue;
@@ -973,10 +974,17 @@ public class AlignmentUtilsTests
   @Test(groups = { "Functional" })
   public void testMakeCdsAlignment()
   {
+    /*
+     * scenario:
+     *     dna1 --> [4, 6] [10,12]        --> pep1 
+     *     dna2 --> [1, 3] [7, 9] [13,15] --> pep1 
+     */
     SequenceI dna1 = new Sequence("dna1", "aaaGGGcccTTTaaa");
     SequenceI dna2 = new Sequence("dna2", "GGGcccTTTaaaCCC");
     SequenceI pep1 = new Sequence("pep1", "GF");
     SequenceI pep2 = new Sequence("pep2", "GFP");
+    pep1.addDBRef(new DBRefEntry("UNIPROT", "0", "pep1"));
+    pep2.addDBRef(new DBRefEntry("UNIPROT", "0", "pep2"));
     dna1.createDatasetSequence();
     dna2.createDatasetSequence();
     pep1.createDatasetSequence();
@@ -984,6 +992,19 @@ public class AlignmentUtilsTests
     AlignmentI dna = new Alignment(new SequenceI[] { dna1, dna2 });
     dna.setDataset(null);
 
+    /*
+     * need a sourceDbRef if we are to construct dbrefs to the CDS
+     * sequence
+     */
+    DBRefEntry dbref = new DBRefEntry("ENSEMBL", "0", "dna1");
+    dna1.getDatasetSequence().setSourceDBRef(dbref);
+    dbref = new DBRefEntry("ENSEMBL", "0", "dna2");
+    dna2.getDatasetSequence().setSourceDBRef(dbref);
+
+    /*
+     * CDS sequences are 'discovered' from dna-to-protein mappings on the alignment
+     * dataset (e.g. added from dbrefs by CrossRef.findXrefSequences)
+     */
     MapList map = new MapList(new int[] { 4, 6, 10, 12 },
             new int[] { 1, 2 }, 3, 1);
     AlignedCodonFrame acf = new AlignedCodonFrame();
@@ -1001,6 +1022,9 @@ public class AlignmentUtilsTests
     AlignmentI cds = AlignmentUtils.makeCdsAlignment(new SequenceI[] {
         dna1, dna2 }, dna.getDataset(), null);
 
+    /*
+     * verify cds sequences
+     */
     assertEquals(2, cds.getSequences().size());
     assertEquals("GGGTTT", cds.getSequenceAt(0).getSequenceAsString());
     assertEquals("GGGTTTCCC", cds.getSequenceAt(1).getSequenceAsString());
@@ -1009,18 +1033,38 @@ public class AlignmentUtilsTests
      * verify shared, extended alignment dataset
      */
     assertSame(dna.getDataset(), cds.getDataset());
-    assertTrue(dna.getDataset().getSequences()
-            .contains(cds.getSequenceAt(0).getDatasetSequence()));
-    assertTrue(dna.getDataset().getSequences()
-            .contains(cds.getSequenceAt(1).getDatasetSequence()));
+    SequenceI cds1Dss = cds.getSequenceAt(0).getDatasetSequence();
+    SequenceI cds2Dss = cds.getSequenceAt(1).getDatasetSequence();
+    assertTrue(dna.getDataset().getSequences().contains(cds1Dss));
+    assertTrue(dna.getDataset().getSequences().contains(cds2Dss));
+
+    /*
+     * verify CDS has a dbref with mapping to peptide
+     */
+    assertNotNull(cds1Dss.getDBRefs());
+    assertEquals(1, cds1Dss.getDBRefs().length);
+    dbref = cds1Dss.getDBRefs()[0];
+    assertEquals("UNIPROT", dbref.getSource());
+    assertEquals("0", dbref.getVersion());
+    assertEquals("pep1", dbref.getAccessionId());
+    assertNotNull(dbref.getMap());
+    assertSame(pep1.getDatasetSequence(), dbref.getMap().getTo());
+    MapList cdsMapping = new MapList(new int[] { 1, 6 },
+            new int[] { 1, 2 }, 3, 1);
+    assertEquals(cdsMapping, dbref.getMap().getMap());
 
     /*
-     * verify cds has dbref with mapping to protein and vice versa
+     * verify peptide has added a dbref with reverse mapping to CDS
      */
-    DBRefEntry[] cdsDbrefs = cds.getSequenceAt(0).getDBRefs();
-    // assertNotNull(cdsDbrefs);
-    // assertEquals(1, cdsDbrefs.length);
-    // assertNotNull(cdsDbrefs[0].getMap());
+    assertNotNull(pep1.getDBRefs());
+    assertEquals(2, pep1.getDBRefs().length);
+    dbref = pep1.getDBRefs()[1];
+    assertEquals("ENSEMBL", dbref.getSource());
+    assertEquals("0", dbref.getVersion());
+    assertEquals("CDS|dna1", dbref.getAccessionId());
+    assertNotNull(dbref.getMap());
+    assertSame(cds1Dss, dbref.getMap().getTo());
+    assertEquals(cdsMapping.getInverse(), dbref.getMap().getMap());
 
     /*
      * Verify mappings from CDS to peptide, cDNA to CDS, and cDNA to peptide
@@ -1054,13 +1098,13 @@ public class AlignmentUtilsTests
     SearchResults sr = MappingUtils.buildSearchResults(pep1, 1, mappings);
     assertEquals(1, sr.getResults().size());
     Match m = sr.getResults().get(0);
-    assertSame(cds.getSequenceAt(0).getDatasetSequence(), m.getSequence());
+    assertSame(cds1Dss, m.getSequence());
     assertEquals(1, m.getStart());
     assertEquals(3, m.getEnd());
     // map F to TTT
     sr = MappingUtils.buildSearchResults(pep1, 2, mappings);
     m = sr.getResults().get(0);
-    assertSame(cds.getSequenceAt(0).getDatasetSequence(), m.getSequence());
+    assertSame(cds1Dss, m.getSequence());
     assertEquals(4, m.getStart());
     assertEquals(6, m.getEnd());
 
@@ -1078,19 +1122,19 @@ public class AlignmentUtilsTests
     sr = MappingUtils.buildSearchResults(pep2, 1, mappings);
     assertEquals(1, sr.getResults().size());
     m = sr.getResults().get(0);
-    assertSame(cds.getSequenceAt(1).getDatasetSequence(), m.getSequence());
+    assertSame(cds2Dss, m.getSequence());
     assertEquals(1, m.getStart());
     assertEquals(3, m.getEnd());
     // map F to TTT
     sr = MappingUtils.buildSearchResults(pep2, 2, mappings);
     m = sr.getResults().get(0);
-    assertSame(cds.getSequenceAt(1).getDatasetSequence(), m.getSequence());
+    assertSame(cds2Dss, m.getSequence());
     assertEquals(4, m.getStart());
     assertEquals(6, m.getEnd());
     // map P to CCC
     sr = MappingUtils.buildSearchResults(pep2, 3, mappings);
     m = sr.getResults().get(0);
-    assertSame(cds.getSequenceAt(1).getDatasetSequence(), m.getSequence());
+    assertSame(cds2Dss, m.getSequence());
     assertEquals(7, m.getStart());
     assertEquals(9, m.getEnd());
   }
@@ -1175,7 +1219,7 @@ public class AlignmentUtilsTests
     SequenceI cdsSeq = cds.get(0);
     assertEquals("GGGTTT", cdsSeq.getSequenceAsString());
     // assertEquals("dna1|A12345", cdsSeq.getName());
-    assertEquals("dna1|pep1", cdsSeq.getName());
+    assertEquals("CDS|dna1", cdsSeq.getName());
     // assertEquals(1, cdsSeq.getDBRefs().length);
     // DBRefEntry cdsRef = cdsSeq.getDBRefs()[0];
     // assertEquals("EMBLCDS", cdsRef.getSource());
@@ -1185,7 +1229,7 @@ public class AlignmentUtilsTests
     cdsSeq = cds.get(1);
     assertEquals("aaaccc", cdsSeq.getSequenceAsString());
     // assertEquals("dna1|A12346", cdsSeq.getName());
-    assertEquals("dna1|pep2", cdsSeq.getName());
+    assertEquals("CDS|dna1", cdsSeq.getName());
     // assertEquals(1, cdsSeq.getDBRefs().length);
     // cdsRef = cdsSeq.getDBRefs()[0];
     // assertEquals("EMBLCDS", cdsRef.getSource());
@@ -1195,7 +1239,7 @@ public class AlignmentUtilsTests
     cdsSeq = cds.get(2);
     assertEquals("aaaTTT", cdsSeq.getSequenceAsString());
     // assertEquals("dna1|A12347", cdsSeq.getName());
-    assertEquals("dna1|pep3", cdsSeq.getName());
+    assertEquals("CDS|dna1", cdsSeq.getName());
     // assertEquals(1, cdsSeq.getDBRefs().length);
     // cdsRef = cdsSeq.getDBRefs()[0];
     // assertEquals("EMBLCDS", cdsRef.getSource());
@@ -2261,7 +2305,7 @@ public class AlignmentUtilsTests
      * execute method under test to find CDS for EMBL peptides only
      */
     AlignmentI cds = AlignmentUtils.makeCdsAlignment(new SequenceI[] {
-        dna1, dna2 }, dna.getDataset(), emblPeptides);
+        dna1, dna2 }, dna.getDataset(), emblPeptides.getSequencesArray());
   
     assertEquals(2, cds.getSequences().size());
     assertEquals("GGGTTT", cds.getSequenceAt(0).getSequenceAsString());