added mode to retrieve single SSS matches only

Actelion · Nov 27, 2023 · 48dd584 · 48dd584
1 parent e9be8d1
commit 48dd584
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 4 deletions.
diff --git a/src/main/java/com/actelion/research/chem/StructureSearch.java b/src/main/java/com/actelion/research/chem/StructureSearch.java
@@ -57,7 +57,7 @@ public class StructureSearch {
 	private volatile StructureSearchDataSource mDataSource;
 	private volatile StructureSearchController mSearchController;
 	private volatile ProgressController mProgressController;
-	private volatile StereoMolecule[] mQueryFragment;
+	private volatile StereoMolecule[] mQueryFragment,mDoubleQueryFragment;
 	private volatile ByteArrayComparator mIDCodeComparator;
 	private volatile DescriptorHandler mDescriptorHandler;
 	private volatile Object[] mQueryDescriptor;
@@ -156,6 +156,15 @@ public int[] start() {
 						mQueryFragment[i] = new IDCodeParser(false).getCompactMolecule(mSpecification.getIDCode(i));
 						mQueryFragment[i].ensureHelperArrays(Molecule.cHelperParities);
 						}
+					if (mSpecification.isSingleMatchOnly()) {
+						mDoubleQueryFragment = new StereoMolecule[queryStructureCount];
+						for (int i=0; i<queryStructureCount; i++) {
+							mDoubleQueryFragment[i] = new StereoMolecule(mQueryFragment[i].getAtoms(), mQueryFragment[i].getBonds());
+							mDoubleQueryFragment[i].addMolecule(mQueryFragment[i]);
+							mDoubleQueryFragment[i].addMolecule(mQueryFragment[i]);
+							mDoubleQueryFragment[i].ensureHelperArrays(Molecule.cHelperParities);
+							}
+						}
 					}
 				else {
 					final String descriptorShortName = mSpecification.getDescriptorShortName();
@@ -310,8 +319,17 @@ public void run() {
 							for (int i=0; i<mQueryFragment.length; i++) {
 								mSSSearcher.setFragment(mQueryFragment[i], (long[])mQueryDescriptor[i]);
 								if (mSSSearcher.isFragmentInMolecule()) {
-									isMatch = true;
-									break;
+									if (mSpecification.isSingleMatchOnly()) {
+										mSSSearcher.setFragment(mDoubleQueryFragment[i], (long[])mQueryDescriptor[i]);
+										if (!mSSSearcher.isFragmentInMolecule()) {
+											isMatch = true;
+											break;
+											}
+										}
+									else {
+										isMatch = true;
+										break;
+										}
 									}
 								}
 							}

diff --git a/src/main/java/com/actelion/research/chem/StructureSearchSpecification.java b/src/main/java/com/actelion/research/chem/StructureSearchSpecification.java
@@ -52,7 +52,8 @@ public class StructureSearchSpecification implements Serializable {
 	public static final int TYPE_TAUTOMER_NO_STEREO	= 0x000006;
 	public static final int TYPE_BACKBONE_NO_STEREO	= 0x000007;
 
-	public static final int MODE_LARGEST_FRAGMENT_ONLY		= 0x000100;
+	public static final int MODE_LARGEST_FRAGMENT_ONLY	= 0x000100;
+	public static final int MODE_SINGLE_MATCH_ONLY = 0x000200;
 
 	private int mSearchType;
 	private byte[][] mIDCode;
@@ -178,16 +179,36 @@ public boolean isLargestFragmentOnly() {
 		return (mSearchType & MODE_LARGEST_FRAGMENT_ONLY) != 0;
 		}
 
+	public boolean isSingleMatchOnly() {
+		return (mSearchType & MODE_SINGLE_MATCH_ONLY) != 0;
+	}
+
 	public void removeDescriptors() {
 		mDescriptor = null;
 		}
 
+	/**
+	 *
+	 * @param b
+	 */
 	public void setLargestFragmentOnly(boolean b) {
 		mSearchType &= ~MODE_LARGEST_FRAGMENT_ONLY;
 		if (b)
 			mSearchType |= MODE_LARGEST_FRAGMENT_ONLY;
 		}
 
+	/**
+	 * In case of a substructure search, as default a molecule is considered a match if the query
+	 * structure is found once or multiple times. To consider only single matches a match, call this
+	 * method with argument true.
+	 * @param b
+	 */
+	public void setSingleMatchOnly(boolean b) {
+		mSearchType &= ~MODE_SINGLE_MATCH_ONLY;
+		if (b)
+			mSearchType |= MODE_SINGLE_MATCH_ONLY;
+		}
+
 	public String getDescriptorShortName() {
 		return mDescriptorShortName;
 		}