fixed a few issues with rule bond query features

src/main/java/com/actelion/research/chem/reaction/mapping/ChemicalRule.java

@@ -71,6 +71,10 @@ public ChemicalRule(String name, String idcode, float panalty) {
 
 	public void initialize() {
 		Reaction rxn = ReactionEncoder.decode(mIDCode, false);
+		if (rxn.getReactants() != 1 || rxn.getProducts() != 1)
+			System.out.println("ERROR: Rule '"+mName+"' doesn't contain exactly one reactant and one product!");
+		if (!rxn.isFragment())
+			System.out.println("ERROR: Rule '"+mName+"' reactant and product are not marked as fragment!");
 		mReactant = rxn.getReactant(0);
 		mProduct = rxn.getProduct(0);
 		mReactant.ensureHelperArrays(Molecule.cHelperNeighbours);
@@ -137,8 +141,28 @@ public void initialize() {
 					}
 				}
 			}
+
+		calculatePenalty();
 		}
 
+	private void calculatePenalty() {
+		MappingScorer scorer = new MappingScorer(mReactant, mProduct);
+		int[] reactantMapNo = new int[mReactant.getAllAtoms()];
+		for (int i=0; i<mReactant.getAllAtoms(); i++)
+			reactantMapNo[i] = mReactant.getAtomMapNo(i);
+		int[] productMapNo = new int[mProduct.getAllAtoms()];
+		for (int i=0; i<mProduct.getAllAtoms(); i++)
+			productMapNo[i] = mProduct.getAtomMapNo(i);
+
+		mPanalty = -scorer.scoreMapping(scorer.createReactantToProductAtomMap(reactantMapNo, productMapNo)) * 0.25f;
+//		mPanalty = -scorer.scoreMapping(scorer.createReactantToProductAtomMap(reactantMapNo, productMapNo)) - 1.5f;	// is a positive value
+
+		// The idea is that when a rule is applied, then the score should be better than
+		// the simple calculated score from bond changes, because we know that we use
+		// reasonable chemistry. How much better, whether a constant of whether rule-dependent
+		// remains to be determined...
+	}
+
 	private boolean isTHParityInversion(int reactantAtom, int[] mapNoToProduct) {
 		boolean inversion = false;
 		if (mReactant.getAtomPi(reactantAtom) == 0) {
@@ -210,12 +234,12 @@ public float getPanalty() {
 	 * scoring redundant mapping graphs, these should be sorted out early. Reasons for
 	 * redundant matches may be:<br>
 	 * - if the rule reactant is one fragment, this may be symmetrical Cn or Dn<br>
-	 * - the rule reactant may contain multiple equivalent fragments, e.g. metathese<br>
-	 * - matching atoms in the real reactant my be symmetrical<br>
+	 * - the rule reactant may contain multiple equivalent fragments, e.g. metathesis<br>
+	 * - matching atoms in the real reactant may be symmetrical<br>
 	 * Otherwise, there certain causes may exist, that break these symmetries:<br>
-	 * - a symmetrical rule fragment must not considered symmetrical, if it doesn't react
+	 * - a symmetrical rule fragment must not be considered symmetrical, if it doesn't react
 	 *   symmetrically, i.e. if its matching rule product atoms are not equivalent anymore.
-	 * - in case of multiple symmetrical fragments in the rule's reactant (e.g. metathese),
+	 * - in case of multiple symmetrical fragments in the rule's reactant (e.g. metathesis),
 	 *   inverted/rotated individual fragment matches cause different products, that means
 	 *   that the relative match orientation of multiple symmetrical fragments breaks symmetry,
 	 *   if the real matching atoms are not equivalent.<br>

src/main/java/com/actelion/research/chem/reaction/mapping/ChemicalRuleEnhancedReactionMapper.java

@@ -48,6 +48,8 @@ public class ChemicalRuleEnhancedReactionMapper implements IReactionMapper {
 	// Of course, when a rule is applied, then only the mapped region of the rule is used as a template to
 	// change bonding of the reaction the rule is applied to. Nevertheless, the rule's entire reactant is used
 	// for the substructure search that identifies applicability.
+	// NOTE: Rule reactions must contain one reactant and one product only (usually these consist of multiple fragments)!
+	// NOTE: You may use the idcodeexplorer.jar to generate new rules.
 	private static final ChemicalRule[] CHEMICAL_RULE = {
 // With cleaned coordinates:
 			new ChemicalRule("d","gGQ@@eKtRA@!gGQ@@djqRA@#qMsT qM\\V#!B_qL@Dw}l@Fp@Dp !B@AL@[@@S@Fp@Dp##", 3.5f),
@@ -70,10 +72,10 @@ public class ChemicalRuleEnhancedReactionMapper implements IReactionMapper {
 			new ChemicalRule("Sakurai", "gOQH@wAINvZ@pdcFe@x@!gOQH@wAIgJi@pdcFe@x@#qreKx qrLkx#!BDpAl@IknDw|S@Fp@ !Bb@JH?_x@b@JH?Ven##", 4.5f),
 			new ChemicalRule("Mitsunobu", "gFP`ATfRhdPp`}KEYg]d@!gFP`ATfRhd`pekL{l`#qrLk qZLn#!B@hc}b@C~@h`YM` !B@hc}b@C~@h`YM`##", 4.5f),
 
-			new ChemicalRule("Aldol-Addition", "gOQ@AdTAcW@[Q^crJTLES`DO}b@!gGQ@@dsuVAcJg@H@#qYEbp qYub#!B@O{|b@Jw\\o{~@Oy? !Bb@K~@Hc}b@JH@`##", 1.5f),
-			new ChemicalRule("Aldol-Condensation", "gOQ@AdTAcW@[Q^crJTLES`DO}b@!gFQ@@`v|pblHHqIpB@#qYEbp q\\VM#!B@O{|b@Jw\\o{~@Oy? !BxpE?[@@S@Fp@Dp##", 2.5f),
-			new ChemicalRule("Acetal-Aldol-Addition", "dmdB@@serQS@sJjnl@p`Xir\\@`j\\@`!daxL@@[df[Zj|@qQdxACdxA@#qB@`OuX qBtM{#!B?[_}b@Jw_?{}mwk~?X`Bmwvw_[\\ !Bb@K~@Hc}b@Jw@h`BmpH##", 1.5f),
-			new ChemicalRule("Acetal-Aldol-Condensation", "dmdB@@serQS@sJjnl@p`Xir\\@`j\\@`!gNp`ITkez^@qTS`DJg@H@#qB@`OuX qOuS`#!B?[_}b@Jw_?{}mwk~?X`Bm?vw_[\\ !Bm?rH__y?b@K~_xa}##", 1.5f),
+			new ChemicalRule("Aldol-Addition", "gOQ@AdTAcS@[Q^crJTLES`DJsL?vH!gGQ@@dsuRAcJg@HUaX#qYEbp qYub#!BxOyBzLKg`dG~xG~{ !Bb@K~@Hc}FBIA@@##", 1.5f),
+			new ChemicalRule("Aldol-Condensation", "gOQ@AdTAcS@[Q^crJTLES`DJsL?vH!gFQ@@`rrpdlHHpipBEXf@#qYEbp q^aU#!B{ZRRqA?AQfyA@L_C !B}QFw@h`B_tnH_P##", 2.5f),
+			new ChemicalRule("Acetal-Aldol-Addition", "dmdB@@serQS@sJjfd@p`Xir\\@`j\\@aUJXK@!daxL@@[df[ZjT@qQdxACdxABjTqf@#qB@`OuX qBtM{#!B_]]}rHKWw?y}uy[~GJbBu{wWqG| !BfJK~TIa]fJJghg{`pP@##", 1.5f),
+			new ChemicalRule("Acetal-Aldol-Condensation", "dmdB@@serQS@sJjfd@p`Xir\\@`j\\@aUJXK@!gNp`CTjUiV@qQS`DJg@HUVXV@#qB@`OuX qqj{`#!B?[_}b@Jw_?{}m~[~[N@Bm?vwkN| !BfsK~yzPrw}m?rzQM##", 1.5f),
 			new ChemicalRule("Acetal-Aldol-Condensation-Cyclization", "dkLB@@ZURYUvUjljh@paHpr\\@`!didD@@yIfUXXHL@CFNS`D@#IXljNPY@@@ IXljIyA#!BbOw~_x`Bm?vH?[_}b@JH?_y?b@Jw?Xc} !BbOvH?Oy??`BH?Xa}?`C~_p##", 7.5f),
 			new ChemicalRule("Enolester-Cleavage", "gOQ`@fdscT`_Qp!gOQ`@cdTASS@P#q}Kr` q}cNP#!B@k]}mpC~@k]}mqdY !Bb@K~@Hc}BfzH@hc}##", 5.3f),
 
@@ -91,7 +93,7 @@ public class ChemicalRuleEnhancedReactionMapper implements IReactionMapper {
 			// two-step
 			new ChemicalRule("Elimination-Claisen", "gNp`AldTQji@~a`!gOP`@teTZdCzN@#qtHUX qtSi@#!Bm?vH?[\\B?g~H@hc} !B@AL@[@@S@Fp@DweA##", 4.5f),
 			new ChemicalRule("imineFormationAzaCope", "daZH@LAIMUjd@pRL@!daZH@HAAn]jd@p`@#IGfaLJ` IFDzfK@#!BDpAl@IkqDpAl@AL@[@@ !BFaFw@h`BbOw~@H`BbOt##", 8.5f),
-			new ChemicalRule("didehydroCopeWithAromatisation", "gNp@DiuVYDsj`rLwA`!gOp@DjWkB@@H#q]yr` q\\oQp#!Bm?vH@k\\Bm?w~_{_} !Bm?w~@Hc}mpJw@ox@##", 4.5f),
+			new ChemicalRule("didehydroCopeWithAromatisation", "gNp@Di]ejDcjcr|wK`!gOp@DjWkB@@H#qrLkx q\\oQp#!B?g~H?K_}bGvH?H`B !Bm?w~@Hc}mpJw@ox@##", 4.5f),
 
 			// multi-step with cyclisation/condensation
 			new ChemicalRule("symAldolNitrogenRing", "dovJ@GBfttf\\v\\qjViPCADGbDodnGp!doNJ@JCSmtefWTCaYjje@H#IlZXi]]yL~C IqMVCzaIim?#!BQtl_riY?Qtl_rfuvNCQ`uZd@NCQ`uVVu}?sA]P !B?`BH@ox@bGvH@k\\Bb@JH_Xa}b@K~_rYltUr|W@##", 0.5f),
@@ -114,6 +116,9 @@ public class ChemicalRuleEnhancedReactionMapper implements IReactionMapper {
 			new ChemicalRule("Epoxydation", "gB``ADcdCB@!gC``AhtUPGtt@#qqb qtQ#!BjW}Y\\YX@ !B?g~w?^Va##", 6.3f),
 			new ChemicalRule("oxydativePropargylAmine13Shift", "gKi@HDEZpLHOQP!gJY@BDeVXB#qMr` qNTh#!BqiXTy{U?mW| !B@Fp@DpAl@AL##", 6.5f),
 			new ChemicalRule("Baeyer-Villiger", "gFQ`@[dTAZ`LHP!gFQ`@jdrMPGtl@#qrak qrlK#!B_?{}mwvHs^FVP@ !BbOvH@oy?bOuQzP##", 7.5f),
+
+			// condensation with ring closure
+			new ChemicalRule("Hantzsch Thiazol", "gOYDGaDDHRTve`H!gKXHL@aJWFe`H#qB`ip qiV`#!B_vq?Dw}lL{y?[G|S !BTqa`FbpX?`@##", 3.5f),
 	};
 
 	private static boolean sInitialized;

src/main/java/com/actelion/research/chem/reaction/mapping/MappingScorer.java

@@ -46,11 +46,11 @@ public class MappingScorer {
 	private static final boolean SCORE_SIMPLE = false;
 	private static final boolean SCORE_HYDROGEN = false;
 
-	private StereoMolecule mReactant,mProduct;
+	private final StereoMolecule mReactant,mProduct;
 
 	/**
 	 * Instantiates a mapping scorer that judges the quality of a mapping by adding penalties for every bond
-	 * being broken, created, or changed. In principle the panelty for any created or broken bond is 2.0,
+	 * being broken, created, or changed. In principle the penalty for any created or broken bond is 2.0,
 	 * and for any changed bond order is 1.0. A change from/to delocalized to/from single or double is considered
 	 * a change. Broken or created bonds at typical break locations, e.g. ester cleavage, get slightly lower
 	 * penalties than 2.0. Changes of implicit hydrogen counts contribute with a factor of 2.0.
@@ -97,6 +97,12 @@ public float scoreMapping(int[] reactantToProductAtom) {
 		for (int rBond=0; rBond<mReactant.getBonds(); rBond++) {
 			int rAtom1 = mReactant.getBondAtom(0, rBond);
 			int rAtom2 = mReactant.getBondAtom(1, rBond);
+
+			if (mReactant.isFragment()
+			 && ((mReactant.getAtomQueryFeatures(rAtom1) & Molecule.cAtomQFExcludeGroup) != 0
+			  || (mReactant.getAtomQueryFeatures(rAtom2) & Molecule.cAtomQFExcludeGroup) != 0))
+				continue;
+
 			int pAtom1 = reactantToProductAtom[rAtom1];
 			int pAtom2 = reactantToProductAtom[rAtom2];
 
@@ -138,6 +144,11 @@ public float scoreMapping(int[] reactantToProductAtom) {
 			}
 
 		for (int pBond=0; pBond<mProduct.getBonds(); pBond++) {
+			if (mProduct.isFragment()
+			 && ((mProduct.getAtomQueryFeatures(mProduct.getBondAtom(0, pBond)) & Molecule.cAtomQFExcludeGroup) != 0
+			  || (mProduct.getAtomQueryFeatures(mProduct.getBondAtom(1, pBond)) & Molecule.cAtomQFExcludeGroup) != 0))
+				continue;
+
 			if (!productBondHandled[pBond]) {
 				penalty += getBondCreateOrBreakPenalty(mProduct, pBond);
 

src/main/java/com/actelion/research/gui/editor/GenericEditorArea.java

@@ -2872,7 +2872,7 @@ private void fireEventLater(EditorEvent e) {
 
 	/**
 	 * Redraws the molecule(s) or the reaction after scaling coordinates.
-	 * Then analyses fragment membership and recreate individual molecules, reaction, or markush structure
+	 * Then analyses fragment membership and recreates individual molecules, reaction, or markush structure
 	 * Then, fires molecule change events with userChange=false, i.e. external change.
 	 */
 	public void moleculeChanged() {