Merge branch 'master' into issue929

josemduarte · josemduarte · commit 7f160ef36fd3 · 2021-09-30T17:12:31.000-07:00
diff --git a/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/scop/TestParsingScopDomainPDB.java b/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/scop/TestParsingScopDomainPDB.java
diff --git a/biojava-modfinder/src/test/java/org/biojava/nbio/protmod/phosphosite/TestAcetylation.java b/biojava-modfinder/src/test/java/org/biojava/nbio/protmod/phosphosite/TestAcetylation.java
@@ -24,6 +24,7 @@
 import org.biojava.nbio.phosphosite.Dataset;
 import org.biojava.nbio.phosphosite.Site;
 import org.junit.Before;
+import org.junit.Ignore;
 import org.junit.Test;
 
 import java.io.File;
@@ -91,6 +92,7 @@ private File getLocalFileName(String phosphoSiteFileLocation){
 	 * Tests that the acetylation file can get parsed without problems.
 	 *
 	 */
+	@Ignore("The file needed from the phosphosite resource is no longer available. See #975")
 	@Test
 	public void testAcetylation() throws IOException {
 
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/contact/StructureInterface.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/contact/StructureInterface.java
@@ -113,6 +113,9 @@ public StructureInterface(
 		this.moleculeIds = new Pair<>(firstMoleculeId,secondMoleculeId);
 		this.contacts = contacts;
 		this.transforms = new Pair<>(firstTransf, secondTransf);
+
+		this.groupAsas1 = new TreeMap<>();
+		this.groupAsas2 = new TreeMap<>();
 	}
 
 	/**
@@ -290,12 +293,8 @@ protected Atom[] getAtomsForAsa(int cofactorSizeToUse) {
 		Atom[] atoms2 = getSecondAtomsForAsa(cofactorSizeToUse);
 
 		Atom[] atoms = new Atom[atoms1.length+atoms2.length];
-		for (int i=0;i<atoms1.length;i++) {
-			atoms[i] = atoms1[i];
-		}
-		for (int i=0;i<atoms2.length;i++) {
-			atoms[i+atoms1.length] = atoms2[i];
-		}
+		System.arraycopy(atoms1, 0, atoms, 0, atoms1.length);
+		System.arraycopy(atoms2, 0, atoms, atoms1.length, atoms2.length);
 
 		return atoms;
 	}
@@ -326,7 +325,7 @@ private static Atom[] getAllNonHAtomArray(Atom[] m, int minSizeHetAtomToInclude)
 			atoms.add(a);
 
 		}
-		return atoms.toArray(new Atom[atoms.size()]);
+		return atoms.toArray(new Atom[0]);
 	}
 
 	/**
@@ -425,11 +424,11 @@ public void setFirstGroupAsa(GroupAsa groupAsa) {
 		groupAsas1.put(groupAsa.getGroup().getResidueNumber(), groupAsa);
 	}
 
-	void setFirstGroupAsas(Map<ResidueNumber, GroupAsa> firstGroupAsas) {
+	public void setFirstGroupAsas(Map<ResidueNumber, GroupAsa> firstGroupAsas) {
 		this.groupAsas1 = firstGroupAsas;
 	}
 
-	void setSecondGroupAsas(Map<ResidueNumber, GroupAsa> secondGroupAsas) {
+	public void setSecondGroupAsas(Map<ResidueNumber, GroupAsa> secondGroupAsas) {
 		this.groupAsas2 = secondGroupAsas;
 	}
 
@@ -587,8 +586,8 @@ public void setCluster(StructureInterfaceCluster cluster) {
 	}
 
 	/**
-	 * Calculates the contact overlap score between this StructureInterface and
-	 * the given one. The calculation assumes that both interfaces come from the same structure. The ouput
+	 * Calculates the Jaccard contact set score (intersection over union) between this StructureInterface and
+	 * the given one. The calculation assumes that both interfaces come from the same structure. The output
 	 * will not necessarily make sense if the two interfaces come from different structures.
 	 * The two sides of the given StructureInterface need to match this StructureInterface
 	 * in the sense that they must come from the same Entity, i.e.
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/contact/StructureInterfaceList.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/contact/StructureInterfaceList.java
@@ -365,7 +365,7 @@ public List<StructureInterfaceCluster> getClusters() {
 
 	/**
 	 * Calculate the interface clusters for this StructureInterfaceList
-	 * using a contact overlap score to measure the similarity of interfaces.
+	 * using Jaccard contact set scores to measure the similarity of interfaces.
 	 * Subsequent calls will use the cached value without recomputing the clusters.
 	 * The clusters will be assigned ids by sorting descending by {@link StructureInterfaceCluster#getTotalArea()}
 	 * @param contactOverlapScoreClusterCutoff the contact overlap score above which a pair will be
@@ -382,6 +382,7 @@ public List<StructureInterfaceCluster> getClusters(double contactOverlapScoreClu
 		// nothing to do if we have no interfaces
 		if (list.size()==0) return clusters;
 
+		logger.debug("Calculating all-vs-all Jaccard scores for {} interfaces", list.size());
 		double[][] matrix = new double[list.size()][list.size()];
 
 		for (int i=0;i<list.size();i++) {
@@ -399,6 +400,7 @@ public List<StructureInterfaceCluster> getClusters(double contactOverlapScoreClu
 
 		}
 
+		logger.debug("Will now cluster {} interfaces based on full all-vs-all Jaccard scores matrix", list.size());
 		SingleLinkageClusterer slc = new SingleLinkageClusterer(matrix, true);
 
 		Map<Integer, Set<Integer>> clusteredIndices = slc.getClusters(contactOverlapScoreClusterCutoff);
@@ -434,11 +436,11 @@ public List<StructureInterfaceCluster> getClusters(double contactOverlapScoreClu
 				interf.setCluster(cluster);
 			}
 		}
+		logger.debug("Done clustering {} interfaces based on full all-vs-all Jaccard scores matrix. Found a total of {} clusters", list.size(), clusters.size());
 
 		// now we sort by areas (descending) and assign ids based on that sorting
-		clusters.sort((o1, o2) -> {
-			return Double.compare(o2.getTotalArea(), o1.getTotalArea()); //note we invert so that sorting is descending
-		});
+		clusters.sort((o1, o2) -> Double.compare(o2.getTotalArea(), o1.getTotalArea())); //note we invert so that sorting is descending
+
 		int id = 1;
 		for (StructureInterfaceCluster cluster:clusters) {
 			cluster.setId(id);
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/scop/ScopFactory.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/scop/ScopFactory.java
@@ -51,6 +51,7 @@ public class ScopFactory {
 	private static final Logger logger = LoggerFactory.getLogger(ScopFactory.class);
 
 	// berkeley 2
+	public static final String VERSION_2_0_8 = "2.08";
 	public static final String VERSION_2_0_7 = "2.07";
 	public static final String VERSION_2_0_6 = "2.06";
 	public static final String VERSION_2_0_5 = "2.05";
@@ -78,7 +79,7 @@ public class ScopFactory {
 	public static final String VERSION_1_55 = "1.55";
 
 	// The most recent version as of compilation time
-	public static final String LATEST_VERSION = VERSION_2_0_7;
+	public static final String LATEST_VERSION = VERSION_2_0_8;
 
 	// Hold one instance for each version
 	private static Map<String,ScopDatabase> versionedScopDBs = new HashMap<>();
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/scop/ScopInstallation.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/scop/ScopInstallation.java
@@ -58,7 +58,7 @@
  */
 public class ScopInstallation implements LocalScopDatabase {
 
-	public static final String DEFAULT_VERSION = "1.75";
+	public static final String DEFAULT_VERSION = ScopFactory.LATEST_VERSION;
 
 	private static final Logger logger = LoggerFactory.getLogger(ScopInstallation.class);
 
@@ -354,7 +354,7 @@ public ScopDomain getDomainByScopID(String scopId) {
 		if ( scopId.length() < 6) {
 			throw new ScopIOException("Does not look like a scop ID! " + scopId);
 		}
-		String pdbId = scopId.substring(1,5);
+		String pdbId = scopId.substring(1,5); //TODO handle this when you handle extended PDBId (PDB ID)
 		List<ScopDomain> doms = getDomainsForPDB(pdbId);
 		if ( doms == null)
 			return null;
diff --git a/biojava-structure/src/test/java/org/biojava/nbio/structure/align/util/AtomCacheTest.java b/biojava-structure/src/test/java/org/biojava/nbio/structure/align/util/AtomCacheTest.java
@@ -64,6 +64,7 @@
 import org.biojava.nbio.structure.test.util.GlobalsHelper;
 import org.junit.After;
 import org.junit.Before;
+import org.junit.Ignore;
 import org.junit.Test;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -88,7 +89,7 @@ public void setUp() {
 		StructureIO.setAtomCache(cache);
 
 		// Use a fixed SCOP version for stability
-		ScopFactory.setScopDatabase(ScopFactory.VERSION_1_75B);
+		ScopFactory.setScopDatabase(ScopFactory.LATEST_VERSION);
 	}
 
 	@After
@@ -166,6 +167,7 @@ public void testGetStructureForDomain3() throws IOException, StructureException
 	 * @throws IOException
 	 * @throws StructureException
 	 */
+	@Ignore("SCOP files < 2.08 are no longer available")
 	@Test
 	public void testGetStructureForChainlessDomains() throws IOException, StructureException {
 		ScopDatabase scop = ScopFactory.getSCOP(ScopFactory.VERSION_1_71); // Uses the range '1-135' without a chain
diff --git a/pom.xml b/pom.xml
@@ -457,7 +457,7 @@
 			<dependency>
 				<groupId>org.biojava.thirdparty</groupId>
 				<artifactId>forester</artifactId>
-				<version>1.038</version>
+				<version>1.039</version>
 			</dependency>
 			<dependency>
 			    <groupId>javax.xml.bind</groupId>
