Allow reading of cumulated double bonds from InChI's

cdk · Dec 5, 2021 · 6e01e55 · 6e01e55
1 parent 5471fae
commit 6e01e55
Show file tree

Hide file tree

Showing 3 changed files with 91 additions and 18 deletions.
diff --git a/base/core/src/main/java/org/openscience/cdk/stereo/ExtendedCisTrans.java b/base/core/src/main/java/org/openscience/cdk/stereo/ExtendedCisTrans.java
@@ -28,6 +28,7 @@
 import org.openscience.cdk.interfaces.IBond;
 import org.openscience.cdk.interfaces.IStereoElement;
 
+import java.util.ArrayList;
 import java.util.List;
 
 import static org.openscience.cdk.interfaces.ITetrahedralChirality.Stereo;
@@ -92,6 +93,45 @@ public static IAtom[] findTerminalAtoms(IAtomContainer container, IBond focus) {
         return null;
     }
 
+    /**
+     * Locate the central double-bond in a chain of cumulated double bonds.
+     *
+     * <pre>
+     * A = C = C = B
+     *       ^
+     * A = C = C = C = C = B
+     *           ^
+     * </pre>
+     *
+     * @param mol molecule
+     * @param atom at atom from either end of the cumulated chains
+     * @return the central bond, or null if not found
+     */
+    public static IBond findCentralBond(IAtomContainer mol, IAtom atom) {
+        List<IBond> bonds = new ArrayList<>();
+        IAtom prevAtom = atom;
+        IBond prevBond = null;
+        boolean found;
+        do {
+            found = false;
+            for (IBond bond : mol.getConnectedBondsList(prevAtom)) {
+                if (prevBond == bond)
+                    continue;
+                if (bond.getOrder() == IBond.Order.DOUBLE) {
+                    bonds.add(bond);
+                    found = true;
+                    prevBond = bond;
+                    prevAtom = bond.getOther(prevAtom);
+                    break;
+                }
+            }
+        } while (found);
+        int nbonds = bonds.size();
+        if ((nbonds&0x1) == 0)
+            return null; // is even => false
+        return bonds.get(nbonds/2);
+    }
+
     /**
      * {@inheritDoc}
      */

diff --git a/storage/inchi/src/main/java/org/openscience/cdk/inchi/InChIToStructure.java b/storage/inchi/src/main/java/org/openscience/cdk/inchi/InChIToStructure.java
@@ -47,6 +47,7 @@
 import org.openscience.cdk.interfaces.IStereoElement;
 import org.openscience.cdk.interfaces.ITetrahedralChirality;
 import org.openscience.cdk.stereo.DoubleBondStereochemistry;
+import org.openscience.cdk.stereo.ExtendedCisTrans;
 import org.openscience.cdk.stereo.ExtendedTetrahedral;
 import org.openscience.cdk.tools.periodictable.PeriodicTable;
 
@@ -156,6 +157,14 @@ protected InChIToStructure(String inchi, IChemObjectBuilder builder, List<String
         generateAtomContainerFromInchi(builder);
     }
 
+    /**
+     * Flip the storage order of atoms in a bond.
+     * @param bond the bond
+     */
+    private void flip(IBond bond) {
+        bond.setAtoms(new IAtom[]{bond.getEnd(), bond.getBegin()});
+    }
+
     /**
      * Gets structure from InChI, and converts InChI library data structure
      * into an IAtomContainer.
@@ -355,6 +364,7 @@ else if (stereo == INCHI_BOND_STEREO.SINGLE_1EITHER || stereo == INCHI_BOND_STER
                 assert stereoElement != null;
                 molecule.addStereoElement(stereoElement);
             } else if (stereo0d.getStereoType() == INCHI_STEREOTYPE.DOUBLEBOND) {
+                boolean extended = false;
                 JniInchiAtom[] neighbors = stereo0d.getNeighbors();
 
                 // from JNI InChI doc
@@ -369,28 +379,35 @@ else if (stereo == INCHI_BOND_STEREO.SINGLE_1EITHER || stereo == INCHI_BOND_STER
                 IAtom b = inchiCdkAtomMap.get(neighbors[2]);
                 IAtom y = inchiCdkAtomMap.get(neighbors[3]);
 
-                // XXX: AtomContainer is doing slow lookup
                 IBond stereoBond = molecule.getBond(a, b);
-                stereoBond.setAtoms(new IAtom[]{a, b}); // ensure a is first atom
-
-                Conformation conformation = null;
-
-                switch (stereo0d.getParity()) {
-                    case ODD:
-                        conformation = Conformation.TOGETHER;
-                        break;
-                    case EVEN:
-                        conformation = Conformation.OPPOSITE;
-                        break;
+                if (stereoBond == null) {
+                    extended = true;
+                    IBond tmp = null;
+                    // A = C = C = B
+                    stereoBond = ExtendedCisTrans.findCentralBond(molecule, a);
+                    if (stereoBond == null)
+                        continue; // warn on invalid input?
+                    IAtom[] ends = ExtendedCisTrans.findTerminalAtoms(molecule, stereoBond);
+                    assert ends != null;
+                    if (ends[0] != a)
+                        flip(stereoBond);
+                } else {
+                    flip(stereoBond);
                 }
 
-                // unspecified not stored
-                if (conformation == null) continue;
+                int config = IStereoElement.TOGETHER;
+                if (stereo0d.getParity() == INCHI_PARITY.EVEN)
+                    config = IStereoElement.OPPOSITE;
 
-                molecule.addStereoElement(new DoubleBondStereochemistry(stereoBond, new IBond[]{molecule.getBond(x, a),
-                        molecule.getBond(b, y)}, conformation));
-            } else {
-                // TODO - other types of atom parity - double bond, etc
+                if (extended) {
+                    molecule.addStereoElement(new ExtendedCisTrans(stereoBond,
+                            new IBond[]{molecule.getBond(x, a),
+                                        molecule.getBond(b, y)}, config));
+                } else {
+                    molecule.addStereoElement(new DoubleBondStereochemistry(stereoBond,
+                            new IBond[]{molecule.getBond(x, a),
+                                        molecule.getBond(b, y)}, config));
+                }
             }
         }
     }

diff --git a/storage/inchi/src/test/java/org/openscience/cdk/inchi/InChIToStructureTest.java b/storage/inchi/src/test/java/org/openscience/cdk/inchi/InChIToStructureTest.java
@@ -265,4 +265,20 @@ public void isotope() throws Exception{
         assertThat(mol.getAtomCount(), is(1));
         assertThat(mol.getAtom(0).getMassNumber(), is(121));
     }
+
+    @Test
+    public void testExtendedCisTrans() throws Exception {
+        IAtomContainer mol = InChIGeneratorFactory.getInstance()
+                .getInChIToStructure("InChI=1/C4BrClFI/c5-3(8)1-2-4(6)7/b4-3-",
+                        SilentChemObjectBuilder.getInstance()).getAtomContainer();
+        assertNotNull(mol);
+        int nExtendedCisTrans = 0;
+        for (IStereoElement se : mol.stereoElements()) {
+            if (se.getConfig() != IStereoElement.CU)
+                nExtendedCisTrans++;
+            else
+                Assert.fail("Expected onl extended cis/trans");
+        }
+        Assert.assertEquals(1, nExtendedCisTrans);
+    }
 }