Implement counting fps for 'SubstructureFingerprints'

- Substructures specified via SMARTS are counted using Vento-Foggia algorithm - Test have been implemented
cdk · Jun 24, 2018 · f47df30 · f47df30
1 parent c3a253c
commit f47df30
Show file tree

Hide file tree

Showing 2 changed files with 167 additions and 9 deletions.
diff --git a/.../fingerprint/src/main/java/org/openscience/cdk/fingerprint/SubstructureFingerprinter.java b/.../fingerprint/src/main/java/org/openscience/cdk/fingerprint/SubstructureFingerprinter.java
@@ -24,10 +24,16 @@
 
 import org.openscience.cdk.exception.CDKException;
 import org.openscience.cdk.interfaces.IAtomContainer;
+import org.openscience.cdk.interfaces.IChemObjectBuilder;
+import org.openscience.cdk.isomorphism.Pattern;
+import org.openscience.cdk.isomorphism.VentoFoggia;
+import org.openscience.cdk.isomorphism.matchers.smarts.SmartsMatchers;
 import org.openscience.cdk.smiles.smarts.SMARTSQueryTool;
+import org.openscience.cdk.smiles.smarts.parser.SMARTSParser;
 
 import java.util.BitSet;
 import java.util.Map;
+import java.util.TreeMap;
 
 /**
  * {@link IFingerprinter} that gives a bit set which has a size equal to the number
@@ -397,8 +403,6 @@ public SubstructureFingerprinter(String[] smarts) {
      *     <li>Countable MACCS patterns:</li>
      * </ul>
      *
-     * @todo: Somehow we should handle this exceptions in a way, that it is transparent to the user why 'smarts==null'.
-     * 
      * @param type The desired type of substructures.
      */
     public SubstructureFingerprinter(Type type) {
@@ -440,6 +444,61 @@ public IBitFingerprint getBitFingerprint(IAtomContainer atomContainer) throws CD
         return new BitSetFingerprint(fingerPrint);
     }
 
+
+    /** {@inheritDoc} */
+    @Override
+    public ICountFingerprint getCountFingerprint(IAtomContainer atomContainer) throws CDKException {
+        if (smarts == null) {
+            throw new CDKException("No substructures were defined");
+        }
+
+        IChemObjectBuilder aCBuilder = atomContainer.getBuilder();
+
+        // init SMARTS invariants (connectivity, degree, etc)
+
+        SmartsMatchers.prepare(atomContainer, false);
+
+        final Map<Integer, Integer> map = new TreeMap<Integer, Integer>();
+        for (int i = 0; i < smarts.length; i++) {
+            Pattern pattern = VentoFoggia.findSubstructure(SMARTSParser.parse(smarts[i], aCBuilder));
+            map.put(i, pattern.matchAll(atomContainer).stereochemistry().countUnique());
+        }
+
+        final int sz = map.size();
+        final int[] hash = new int[sz], count = new int[sz];
+        int n = 0;
+        for (int h : map.keySet()) {
+            hash[n] = h;
+            count[n++] = map.get(h);
+        }
+
+        return new ICountFingerprint() {
+            @Override
+            public long size() { return smarts.length; }
+
+            @Override
+            public int numOfPopulatedbins() { return sz; }
+
+            @Override
+            public int getCount(int index) { return count[index]; }
+
+            @Override
+            public int getHash(int index) { return hash[index]; }
+
+            @Override
+            public void merge(ICountFingerprint fp) {}
+
+            @Override
+            public void setBehaveAsBitFingerprint(boolean behaveAsBitFingerprint) {}
+
+            @Override
+            public boolean hasHash(int hash) { return map.containsKey(hash); }
+
+            @Override
+            public int getCountForHash(int hash) { return map.get(hash); }
+        };
+    }
+
     /** {@inheritDoc} */
     @Override
     public Map<String, Integer> getRawFingerprint(IAtomContainer iAtomContainer) throws CDKException {
@@ -463,11 +522,4 @@ public int getSize() {
     public String getSubstructure(int bitIndex) {
         return smarts[bitIndex];
     }
-
-    /** {@inheritDoc} */
-    @Override
-    public ICountFingerprint getCountFingerprint(IAtomContainer container) throws CDKException {
-        throw new UnsupportedOperationException();
-    }
-
 }
diff --git a/...gerprint/src/test/java/org/openscience/cdk/fingerprint/SubstructureFingerprinterTest.java b/...gerprint/src/test/java/org/openscience/cdk/fingerprint/SubstructureFingerprinterTest.java
@@ -178,6 +178,7 @@ public void testCountableMACCSBinary2() throws Exception {
         mol = parser.parseSmiles("C([S](O)(=O)=O)C1=C(C=CC=C1)CCCC[N+](=O)[O-]");
         AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
         Aromaticity.cdkLegacy().apply(mol);
+
         bs = printer.getBitFingerprint(mol).asBitSet();
 
         Assert.assertTrue(bs.get(46));
@@ -209,6 +210,111 @@ public void testCountableMACCSBinary2() throws Exception {
         Assert.assertFalse(bs.get(91));
     }
 
+    @Test
+    public void testGetCountFingerprint() throws Exception {}
+
+    @Test
+    public void testCountableMACCSCount2() throws Exception {
+        SmilesParser parser = new SmilesParser(SilentChemObjectBuilder.getInstance());
+        IFingerprinter printer = new SubstructureFingerprinter(SubstructureFingerprinter.Type.COUNTABLE_MACCS166);
+        IAtomContainer mol;
+        ICountFingerprint cfp;
+
+        // Test molecule 1
+        mol = parser.parseSmiles("C([S](O)(=O)=O)C1=C(C=CC=C1)CCCC[N+](=O)[O-]");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(46) == 2);
+        Assert.assertTrue(cfp.getCountForHash(27) == 1);
+        Assert.assertTrue(cfp.getCountForHash(59) == 2);
+        Assert.assertTrue(cfp.getCountForHash(49) == 1);
+        Assert.assertTrue(cfp.getCountForHash(111) == 1);
+        Assert.assertTrue(cfp.getCountForHash(129) == 3);
+        Assert.assertTrue(cfp.getCountForHash(115) == 2);
+        Assert.assertTrue(cfp.getCountForHash(120) == 3);
+        Assert.assertTrue(cfp.getCountForHash(41) == 3);
+
+        Assert.assertTrue(cfp.getCountForHash(93) == 0);
+        Assert.assertTrue(cfp.getCountForHash(91) == 0);
+        Assert.assertTrue(cfp.getCountForHash(24) == 0);
+
+        // Test molecule 2: Diatrizoic acid
+        mol = parser.parseSmiles("CC(=O)NC1=C(C(=C(C(=C1I)C(=O)O)I)NC(=O)C)I");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(15) == 3);
+        Assert.assertTrue(cfp.getCountForHash(135) == 3);
+        Assert.assertTrue(cfp.getCountForHash(139) == 4);
+        Assert.assertTrue(cfp.getCountForHash(93) == 3);
+        Assert.assertTrue(cfp.getCountForHash(73) == 6);
+
+        Assert.assertTrue(cfp.getCountForHash(91) == 0);
+    }
+
+    @Test
+    public void testCountableMACCSCount_Rings() throws Exception {
+        SmilesParser parser = new SmilesParser(SilentChemObjectBuilder.getInstance());
+        IFingerprinter printer = new SubstructureFingerprinter(SubstructureFingerprinter.Type.COUNTABLE_MACCS166);
+        IAtomContainer mol;
+        ICountFingerprint cfp;
+
+        // Aromatic 6-rings
+        mol = parser.parseSmiles("C1=CC=CC(=C1)CCCC2=CC=CC=C2");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(128) == 2); // 6-ring
+        Assert.assertTrue(cfp.getCountForHash(111) == 2); // aromaticity
+
+        Assert.assertTrue(cfp.getCountForHash(7) == 0); // 7-ring
+        Assert.assertTrue(cfp.getCountForHash(82) == 0); // 5-ring
+
+        // Non-aromatic 6-rings
+        mol = parser.parseSmiles("C1CC(CCC1)CCCCC2CCCCC2");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(128) == 2); // 6-ring
+
+        Assert.assertTrue(cfp.getCountForHash(111) == 0); // aromaticity
+        Assert.assertTrue(cfp.getCountForHash(7) == 0); // 7-ring
+        Assert.assertTrue(cfp.getCountForHash(82) == 0); // 5-ring
+
+        // Aromatic 6-ring, 3-ring and 4-ring
+        mol = parser.parseSmiles("C1CC1C(CCC2CCC2)CC3=CC=CC=C3");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(128) == 1); // 6-ring
+        Assert.assertTrue(cfp.getCountForHash(111) == 1); // aromaticity
+        Assert.assertTrue(cfp.getCountForHash(10) == 1); // 3-ring
+        Assert.assertTrue(cfp.getCountForHash(1) == 1); // 4-ring
+
+        Assert.assertTrue(cfp.getCountForHash(7) == 0); // 7-ring
+        Assert.assertTrue(cfp.getCountForHash(82) == 0); // 5-ring
+
+        // Aromatic 6-ring, 3-ring and 4-ring
+        mol = parser.parseSmiles("C1(CC1C(CCC2CCC2)CC3=CC=CC=C3)C(C(C(C4CC4)C5CC5)C6CC6)C7CC7");
+        AtomContainerManipulator.percieveAtomTypesAndConfigureAtoms(mol);
+        Aromaticity.cdkLegacy().apply(mol);
+        cfp = printer.getCountFingerprint(mol);
+
+        Assert.assertTrue(cfp.getCountForHash(128) == 1); // 6-ring
+        Assert.assertTrue(cfp.getCountForHash(111) == 1); // aromaticity
+        Assert.assertTrue(cfp.getCountForHash(10) == 5); // 3-ring
+        Assert.assertTrue(cfp.getCountForHash(1) == 1); // 4-ring
+
+        Assert.assertTrue(cfp.getCountForHash(7) == 0); // 7-ring
+        Assert.assertTrue(cfp.getCountForHash(82) == 0); // 5-ring
+    }
+
     @Test
     public void testCountableMACCSBinary_Rings() throws Exception {
         SmilesParser parser = new SmilesParser(SilentChemObjectBuilder.getInstance());