Merge remote-tracking branch 'origin/master'

pom.xml

@@ -8,7 +8,7 @@
     Please follow the naming scheme YEAR.MONTH.RELEASE_NO_OF_MONTH
     (eg. 2016.4.1 for second release in Apr 2016)
     -->
-    <version>2024.9.4-SNAPSHOT</version>
+    <version>2024.10.1-SNAPSHOT</version>
 
     <name>OpenChemLib</name>
     <description>Open Source Chemistry Library</description>

src/main/java/com/actelion/research/chem/descriptor/flexophore/DistHistHelper.java

@@ -36,6 +36,7 @@
 
 import com.actelion.research.calc.ArrayUtilsCalc;
 import com.actelion.research.calc.statistics.median.ModelMedianInteger;
+import com.actelion.research.chem.descriptor.flexophore.generator.ConstantsFlexophoreGenerator;
 import com.actelion.research.util.datamodel.IntArray;
 
 /**
@@ -174,6 +175,53 @@ public static RangeStatistics getRangeStatistics(MolDistHist mdh){
         return rangeStatisticsTotal;
     }
 
+    public static int count(byte [] a){
+        int c = 0;
+        for (int i = 0; i < a.length; i++) {
+            c += a[i];
+        }
+        return c;
+    }
+
+    public static byte [] normalize(double [] arrHistRaw){
+
+        int countValuesInHistogram = 0;
+        for (int i = 0; i < arrHistRaw.length; i++) {
+            countValuesInHistogram += arrHistRaw[i];
+        }
+        byte [] arrHistPercent = new byte [arrHistRaw.length];
+
+        for (int i = 0; i < arrHistRaw.length; i++) {
+            arrHistPercent[i]= (byte)  (((arrHistRaw[i] / countValuesInHistogram) * ConstantsFlexophoreGenerator.SUM_VAL_HIST) + 0.5);
+        }
+
+//            System.out.println(StringFunctions.toString(arrHistPercent));
+
+        return arrHistPercent;
+
+    }
+    public static byte [] normalize(byte [] arrHistRaw){
+
+        double countValuesInHistogram = 0;
+        for (int i = 0; i < arrHistRaw.length; i++) {
+            countValuesInHistogram += arrHistRaw[i];
+        }
+        byte [] arrHistPercent = new byte [arrHistRaw.length];
+
+        for (int i = 0; i < arrHistRaw.length; i++) {
+
+            double v = ((arrHistRaw[i] / countValuesInHistogram) * ConstantsFlexophoreGenerator.SUM_VAL_HIST) + 0.5;
+
+
+            arrHistPercent[i]= (byte)  v;
+        }
+
+//            System.out.println(StringFunctions.toString(arrHistPercent));
+
+        return arrHistPercent;
+
+    }
+
 
     public static class RangeStatistics {
 

src/main/java/com/actelion/research/chem/descriptor/flexophore/MolDistHist.java

@@ -129,24 +129,16 @@ public void copy(MolDistHist copy){
 	}
 
 	public boolean equals(Object o) {
-		boolean bEQ=true;
+		if(!equalNodes(o)){
+			return false;
+		}
 		MolDistHist mdh=null;
 		try {
 			mdh = (MolDistHist)o;
 		} catch (RuntimeException e) {
 			return false;
 		}
-		if(getNumPPNodes() != mdh.getNumPPNodes())
-			return false;
-
-		for (int i = 0; i < getNumPPNodes(); i++) {
-			PPNode n1 = getNode(i);
-			PPNode n2 = mdh.getNode(i);
-			if(!n1.equals(n2)){
-				bEQ = false;
-				break;
-			}
-		}
+		boolean bEQ=true;
 
 		for (int i = 0; i < getNumPPNodes(); i++) {
 			for (int j = i+1; j < getNumPPNodes(); j++) {
@@ -163,6 +155,28 @@ public boolean equals(Object o) {
 
 		return bEQ;
 	}
+	public boolean equalNodes(Object o) {
+		boolean bEQ=true;
+		MolDistHist mdh=null;
+		try {
+			mdh = (MolDistHist)o;
+		} catch (RuntimeException e) {
+			return false;
+		}
+		if(getNumPPNodes() != mdh.getNumPPNodes())
+			return false;
+
+		for (int i = 0; i < getNumPPNodes(); i++) {
+			PPNode n1 = getNode(i);
+			PPNode n2 = mdh.getNode(i);
+			if(!n1.equals(n2)){
+				bEQ = false;
+				break;
+			}
+		}
+
+		return bEQ;
+	}
 
 	public byte getModeFlexophore() {
 		return modeFlexophore;

src/main/java/com/actelion/research/chem/descriptor/flexophore/MolDistHistHelper.java

@@ -1,5 +1,6 @@
 package com.actelion.research.chem.descriptor.flexophore;
 
+import com.actelion.research.chem.descriptor.flexophore.generator.ConstantsFlexophoreGenerator;
 import com.actelion.research.chem.descriptor.flexophore.generator.SubFlexophoreGenerator;
 import com.actelion.research.util.ArrayUtils;
 import com.actelion.research.util.ByteArray;
@@ -258,4 +259,20 @@ public static MolDistHist createFromNodes (Collection<PPNode> col){
         return a;
     }
 
+    public static void reNormalizeDistHist(MolDistHist mdh, int margin){
+        int n = mdh.getNumPPNodes();
+        for (int i = 0; i < n; i++) {
+            for (int j = i+1; j < n; j++) {
+                byte [] b = mdh.getDistHist(i,j);
+                int c = DistHistHelper.count(b);
+                if(Math.abs(ConstantsFlexophoreGenerator.SUM_VAL_HIST-c) > margin){
+                    byte [] distHistNew = DistHistHelper.normalize(b);
+                    mdh.setDistHist(i,j,distHistNew);
+//                    int c2 = DistHistHelper.count(distHistNew);
+//                    System.out.println(c2);
+                }
+            }
+        }
+    }
+
 }

src/main/java/com/actelion/research/chem/descriptor/flexophore/generator/ConstantsFlexophoreGenerator.java

@@ -4,7 +4,7 @@
 
 public class ConstantsFlexophoreGenerator {
 
-
+    public static final double SUM_VAL_HIST = 100.0;
     /**
      * Defines the resolution for the range. This is also the length for the DistHist vector.
      */

src/main/java/com/actelion/research/chem/descriptor/flexophore/generator/MultCoordFragIndex.java

@@ -3,6 +3,7 @@
 import com.actelion.research.calc.Matrix;
 import com.actelion.research.calc.histogram.MatrixBasedHistogram;
 import com.actelion.research.chem.Coordinates;
+import com.actelion.research.chem.descriptor.flexophore.DistHistHelper;
 import com.actelion.research.chem.descriptor.flexophore.redgraph.SubGraphIndices;
 import com.actelion.research.util.Formatter;
 import com.actelion.research.util.datamodel.DoubleArray;
@@ -80,27 +81,12 @@ public int[] getArrIndexFrag() {
 
         Matrix maHist =  MatrixBasedHistogram.getHistogram(daDistances.get(), maBins);
 
-        double [] arrHist = maHist.getRow(2);
+        double [] arrHistRaw = maHist.getRow(2);
 
 //            System.out.println(StringFunctions.toString(daDistances.get()));
 //            System.out.println(StringFunctions.toString(arrHist));
 
-        int countValuesInHistogram = 0;
-
-        for (int i = 0; i < arrHist.length; i++) {
-            countValuesInHistogram += arrHist[i];
-        }
-
-        // Here, the percentage values for the histograms are calculated.
-        byte [] arrHistPercent = new byte [maHist.getColDim()];
-
-        for (int i = 0; i < arrHist.length; i++) {
-            arrHistPercent[i]= (byte)  (((arrHist[i] / countValuesInHistogram) * 100.0) + 0.5);
-        }
-
-//            System.out.println(StringFunctions.toString(arrHistPercent));
-
-        return arrHistPercent;
+        return DistHistHelper.normalize(arrHistRaw);
 
     }
 

src/main/java/com/actelion/research/util/graph/complete/CompleteGraphMatcher.java

@@ -52,6 +52,7 @@ public class CompleteGraphMatcher<T extends ICompleteGraph> {
 	public static final boolean DEBUG = false;
 
 	// We need at least two pharmacophore points.
+	public static final double TINY_SIM = 0.000001;
 	public static final int MIN_NUM_NODES_SIM = 2;
 
 	public static final int MAX_NUM_NODES = 127;
@@ -82,7 +83,8 @@ public class CompleteGraphMatcher<T extends ICompleteGraph> {
 	private byte [] arrIndexQueryTmp;
 
 	private SolutionCompleteGraph solutionBest;
-
+	private List<SolutionCompleteGraph> solutionsTop;
+
 	private long validSolutions;
 
 	private long createdSolutions;
@@ -129,7 +131,7 @@ private void init(){
 		arrIndexQueryTmp = new byte [MAX_NUM_NODES];
 
 		solutionBest = new SolutionCompleteGraph();
-
+		solutionsTop = new ArrayList<>();
 		nodeSimilarityWithoutSizeDifference = false;
 	}
 
@@ -172,6 +174,7 @@ private void initSearch(){
 		}
 
 		solutionBest = new SolutionCompleteGraph();
+		solutionsTop.clear();
 
 		// System.out.println(liSolution.size());
 	}
@@ -197,11 +200,21 @@ public double calculateSimilarity () {
 			double simMax=0;
 			for (SolutionCompleteGraph solutionCompleteGraph : liSolution) {
 				double sim = objectiveCompleteGraph.getSimilarity(solutionCompleteGraph);
+				solutionCompleteGraph.setSimilarity(sim);
 				if(sim>simMax){
 					simMax=sim;
 					solutionBest=solutionCompleteGraph;
 				}
 			}
+
+			solutionsTop.clear();
+			double simMaxMargin=simMax-TINY_SIM;
+			for (SolutionCompleteGraph scg : liSolution) {
+				if(scg.getSimilarity()>simMaxMargin){
+					solutionsTop.add(scg);
+				}
+			}
+
 			return simMax;
 		}
 
@@ -267,10 +280,22 @@ public double calculateSimilarity () {
 		}
 
 		double similarity = solutionBest.getSimilarity();
-
+
+		solutionsTop.clear();
+		double simMaxMargin=similarity-TINY_SIM;
+		for (SolutionCompleteGraph scg : li) {
+			if(scg.getSimilarity()>simMaxMargin){
+				solutionsTop.add(scg);
+			}
+		}
+
 		return similarity;
 	}
 
+	public List<SolutionCompleteGraph> getSolutionsTop() {
+		return solutionsTop;
+	}
+
 	/**
 	 * Be careful! The histogram similarity is still considered if not explicitly set to false in the objective.
 	 * @return

src/main/java/com/actelion/research/util/graph/complete/SolutionCompleteGraph.java

@@ -135,13 +135,12 @@ public byte getIndexCorrespondingBaseNode(int indexQueryNode) {
 	/**
 	 * The index is the index of the node in the query molecule.
 	 * The value at 'index' is the index of the node in the base molecule.
+	 * Can contain -1 if a node is not mapping.
 	 */
 	public byte [] getSolution (){
 		return arrSolution;
 	}
-
 
-
 	public boolean equals(Object obj) {
 
 		boolean eq = true;