JAL-2029 many-to-many EnsemblCDS-to-Uniprot mappings

[jalview.git] / test / jalview / datamodel / xdb / embl / EmblEntryTest.java

diff --git a/test/jalview/datamodel/xdb/embl/EmblEntryTest.java b/test/jalview/datamodel/xdb/embl/EmblEntryTest.java
new file mode 100644 (file)
index 0000000..9fffc45
--- /dev/null
+++ b/test/jalview/datamodel/xdb/embl/EmblEntryTest.java
@@ -0,0 +1,308 @@
+package jalview.datamodel.xdb.embl;
+
+import static org.testng.AssertJUnit.assertEquals;
+import static org.testng.AssertJUnit.assertSame;
+
+import jalview.util.MappingUtils;
+
+import java.util.Arrays;
+import java.util.Vector;
+
+import org.testng.annotations.Test;
+
+public class EmblEntryTest
+{
+  @Test(groups = "Functional")
+  public void testGetCdsRanges()
+  {
+    EmblEntry testee = new EmblEntry();
+
+    /*
+     * Make a (CDS) Feature with 4 locations
+     */
+    EmblFeature cds = new EmblFeature();
+    Vector<EmblFeatureLocations> locs = new Vector<EmblFeatureLocations>();
+    cds.setLocations(locs);
+
+    /*
+     * single range [10-20]
+     */
+    EmblFeatureLocations loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    Vector<EmblFeatureLocElement> elements = new Vector<EmblFeatureLocElement>();
+    EmblFeatureLocElement locElement = new EmblFeatureLocElement();
+    BasePosition b1 = new BasePosition();
+    b1.setPos("10");
+    BasePosition b2 = new BasePosition();
+    b2.setPos("20");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * complement range [30-40]
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(true);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("30");
+    b2 = new BasePosition();
+    b2.setPos("40");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * join range [50-60], [70-80]
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("join");
+    loc.setLocationComplement(false);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("50");
+    b2 = new BasePosition();
+    b2.setPos("60");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("70");
+    b2 = new BasePosition();
+    b2.setPos("80");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * complement range [90-100], [110-120]
+     * this should be the same as complement(join(90..100,110.120))
+     * which is "join 90-100 and 110-120, then complement"
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("join");
+    loc.setLocationComplement(true);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("90");
+    b2 = new BasePosition();
+    b2.setPos("100");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("110");
+    b2 = new BasePosition();
+    b2.setPos("120");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    int[] exons = testee.getCdsRanges(cds);
+    assertEquals("[10, 20, 40, 30, 50, 60, 70, 80, 120, 110, 100, 90]",
+            Arrays.toString(exons));
+  }
+
+  @Test(groups = "Functional")
+  public void testGetCdsRanges_badData()
+  {
+    EmblEntry testee = new EmblEntry();
+
+    /*
+     * Make a (CDS) Feature with 4 locations
+     */
+    EmblFeature cds = new EmblFeature();
+    Vector<EmblFeatureLocations> locs = new Vector<EmblFeatureLocations>();
+    cds.setLocations(locs);
+
+    /*
+     * single range [10-20]
+     */
+    EmblFeatureLocations loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    Vector<EmblFeatureLocElement> elements = new Vector<EmblFeatureLocElement>();
+    EmblFeatureLocElement locElement = new EmblFeatureLocElement();
+    BasePosition b1 = new BasePosition();
+    b1.setPos("10");
+    BasePosition b2 = new BasePosition();
+    b2.setPos("20");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * single range with missing end position - should be skipped
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("30");
+    locElement.setBasePositions(new BasePosition[] { b1 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * single range with extra base position - should be skipped
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("30");
+    locElement.setBasePositions(new BasePosition[] { b1, b1, b1 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    /*
+     * single valid range [50-60] to finish
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("50");
+    b2 = new BasePosition();
+    b2.setPos("60");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+
+    int[] exons = testee.getCdsRanges(cds);
+    assertEquals("[10, 20, 50, 60]", Arrays.toString(exons));
+  }
+
+  /**
+   * Test retrieval of exon locations matching an accession id
+   */
+  @Test(groups = "Functional")
+  public void testGetCdsRanges_forAccession()
+  {
+    EmblEntry testee = new EmblEntry();
+    String accession = "A1234";
+    testee.setAccession(accession);
+    /*
+     * Make a (CDS) Feature with 4 locations
+     */
+    EmblFeature cds = new EmblFeature();
+    Vector<EmblFeatureLocations> locs = new Vector<EmblFeatureLocations>();
+    cds.setLocations(locs);
+  
+    /*
+     * single range [10-20] for 'this' accession
+     */
+    EmblFeatureLocations loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(false);
+    Vector<EmblFeatureLocElement> elements = new Vector<EmblFeatureLocElement>();
+    EmblFeatureLocElement locElement = new EmblFeatureLocElement();
+    locElement.setAccession(accession);
+    BasePosition b1 = new BasePosition();
+    b1.setPos("10");
+    BasePosition b2 = new BasePosition();
+    b2.setPos("20");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+  
+    /*
+     * complement range [30-40] - no accession
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("single");
+    loc.setLocationComplement(true);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    b1 = new BasePosition();
+    b1.setPos("30");
+    b2 = new BasePosition();
+    b2.setPos("40");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+  
+    /*
+     * join range [50-60] this accession, [70-80] another
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("join");
+    loc.setLocationComplement(false);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    locElement.setAccession(accession);
+    b1 = new BasePosition();
+    b1.setPos("50");
+    b2 = new BasePosition();
+    b2.setPos("60");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    locElement = new EmblFeatureLocElement();
+    locElement.setAccession("notme");
+    b1 = new BasePosition();
+    b1.setPos("70");
+    b2 = new BasePosition();
+    b2.setPos("80");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+  
+    /*
+     * complement range [90-100] wrong accession, [110-120] good 
+     * this should be the same as complement(join(90..100,110.120))
+     * which is "join 90-100 and 110-120, then complement"
+     */
+    loc = new EmblFeatureLocations();
+    loc.setLocationType("join");
+    loc.setLocationComplement(true);
+    elements = new Vector<EmblFeatureLocElement>();
+    locElement = new EmblFeatureLocElement();
+    locElement.setAccession("wrong");
+    b1 = new BasePosition();
+    b1.setPos("90");
+    b2 = new BasePosition();
+    b2.setPos("100");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    locElement = new EmblFeatureLocElement();
+    locElement.setAccession(accession);
+    b1 = new BasePosition();
+    b1.setPos("110");
+    b2 = new BasePosition();
+    b2.setPos("120");
+    locElement.setBasePositions(new BasePosition[] { b1, b2 });
+    elements.add(locElement);
+    loc.setLocElements(elements);
+    locs.add(loc);
+  
+    /*
+     * verify we pick out only ranges for A1234
+     */
+    int[] exons = testee.getCdsRanges(cds);
+    assertEquals("[10, 20, 50, 60, 120, 110]",
+            Arrays.toString(exons));
+  }
+}