Now identical clustering performed on 95% identical sequences

josemduarte · josemduarte · commit 730d6b467656 · 2017-10-10T17:23:53.000-07:00
diff --git a/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/symmetry/TestQuatSymmetryDetectorExamples.java b/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/symmetry/TestQuatSymmetryDetectorExamples.java
@@ -234,4 +234,27 @@ public void testHelicalLocal() throws IOException, StructureException {
 		assertEquals("A5", results.get(0).getStoichiometry());
 
 	}
+	
+	/**
+	 * A structure with very similar entities (clustering at 95% seq id): 4DZ8
+	 * @throws IOException
+	 * @throws StructureException
+	 */
+	@Test
+	public void testPseudoIdentity95() throws IOException, StructureException {
+		Structure pdb = StructureIO.getStructure("BIO:4DZ8:1");
+
+		SubunitClustererParameters cp = new SubunitClustererParameters();
+		cp.setClustererMethod(SubunitClustererMethod.IDENTITY);
+		QuatSymmetryParameters symmParams = new QuatSymmetryParameters();
+
+		QuatSymmetryResults symmetry = QuatSymmetryDetector.calcGlobalSymmetry(
+				pdb, symmParams, cp);
+
+		assertEquals("C2", symmetry.getSymmetry());
+		assertEquals("A2", symmetry.getStoichiometry());
+		assertFalse(symmetry.isPseudoStoichiometric());
+		assertEquals(SubunitClustererMethod.IDENTITY, symmetry.getSubunitClusters().get(0).getClustererMethod());
+		
+	}
 }
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/cluster/SubunitCluster.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/cluster/SubunitCluster.java
@@ -143,6 +143,29 @@ public boolean mergeIdentical(SubunitCluster other) {
 		return true;
 	}
 
+	/**
+	 * Merges the other SubunitCluster into this one if their representatives
+	 * sequences are more than 0.95 identical on 0.95 of coverage.
+	 * <p>
+	 * The sequence alignment is performed using Smith Waterman, default linear
+	 * {@link SimpleGapPenalty} and BLOSUM62 as scoring matrix.
+	 * 
+	 * @param other
+	 *            SubunitCluster
+	 * @return true if the SubunitClusters were merged, false otherwise
+	 * @throws CompoundNotFoundException
+	 */
+	public boolean mergeIdentity95(SubunitCluster other) throws CompoundNotFoundException {
+		boolean merged = mergeSequence(other, 0.95, 0.95,
+				PairwiseSequenceAlignerType.LOCAL, new SimpleGapPenalty(),
+				SubstitutionMatrixHelper.getBlosum62());
+		
+		if (merged) {
+			this.method = SubunitClustererMethod.IDENTITY;
+		}
+		return merged;
+	}
+
 	/**
 	 * Merges the other SubunitCluster into this one if their representatives
 	 * sequences are similar (higher sequence identity and coverage than the
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/cluster/SubunitClusterer.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/cluster/SubunitClusterer.java
@@ -68,16 +68,22 @@ public static List<SubunitCluster> cluster(List<Subunit> subunits,
 		for (Subunit s : subunits)
 			clusters.add(new SubunitCluster(s));
 
-		// Now merge clusters by IDENTITY
+		// Now merge clusters by 95% IDENTITY
 		for (int c1 = 0; c1 < clusters.size(); c1++) {
 			for (int c2 = clusters.size() - 1; c2 > c1; c2--) {
-				if (clusters.get(c1).mergeIdentical(clusters.get(c2)))
-					clusters.remove(c2);
+				try {
+					if (clusters.get(c1).mergeIdentity95(clusters.get(c2)))
+						clusters.remove(c2);
+				} catch (CompoundNotFoundException e) {
+					logger.warn("Could not merge by Identity95. {}",
+							e.getMessage());
+				}
 			}
 		}
 
 		if (params.getClustererMethod() == SubunitClustererMethod.IDENTITY)
 			return clusters;
+		
 
 		// Now merge clusters by SEQUENCE similarity
 		for (int c1 = 0; c1 < clusters.size(); c1++) {
