More parsing fixes, biojava#305

josemduarte · josemduarte · commit 6f709099edf7 · 2015-07-27T11:51:33.000+02:00
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/Compound.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/Compound.java
@@ -510,6 +510,8 @@ private void initResSerialsMap(Chain c) {
 			logger.warn("No SEQRES groups found in chain {}, will use residue numbers as given (no insertion codes, not necessarily aligned). "
 					+ "Make sure your structure has SEQRES records and that you use FileParsingParameters.setAlignSeqRes(true)", 
 					c.getChainID());
+			// we add a explicit null to the map so that we flag it as unavailable for this chain
+			chains2pdbResNums2ResSerials.put(c.getChainID(), null);
 			return;
 		}
 		
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/StructureTools.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/StructureTools.java
@@ -1718,4 +1718,17 @@ public static boolean isChainWaterOnly(Chain c) {
 		return waterOnly;
 	}
 
+	/**
+	 * Returns true if the given chain is composed of non-polymeric groups only
+	 * @param c
+	 * @return
+	 */
+	public static boolean isChainPureNonPolymer(Chain c) {
+		
+		for (Group g: c.getAtomGroups()) {
+			if (g.getType() == GroupType.AMINOACID || g.getType() == GroupType.NUCLEOTIDE) return false;
+
+		}
+		return true;
+	}
 }
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/io/PDBFileParser.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/io/PDBFileParser.java
@@ -2139,6 +2139,10 @@ private void pdb_MODEL_Handler(String line) {
 					logger.warn("Chain {} ({} atom groups) is composed of water molecules only. Removing it.", 
 							c.getChainID(), c.getAtomGroups().size());
 					it.remove();
+				} else if (StructureTools.isChainPureNonPolymer(c)) {
+					logger.warn("Chain {} ({} atom groups) is composed of non-polymer molecules only. Removing it.", 
+							c.getChainID(), c.getAtomGroups().size());					
+					it.remove();
 				}
 			}
 			structure.addModel(current_model);
@@ -2922,6 +2926,10 @@ private void triggerEndFileChecks(){
 				logger.warn("Chain {} ({} atom groups) is composed of water molecules only. Removing it.", 
 						c.getChainID(), c.getAtomGroups().size());
 				it.remove();
+			} else if (StructureTools.isChainPureNonPolymer(c)) {
+				logger.warn("Chain {} ({} atom groups) is composed of non-polymer molecules only. Removing it.", 
+						c.getChainID(), c.getAtomGroups().size());
+				it.remove();				
 			}
 		}
 		structure.addModel(current_model);
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/io/mmcif/SimpleMMcifConsumer.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/io/mmcif/SimpleMMcifConsumer.java
@@ -748,14 +748,12 @@ public void documentEnd() {
 				it = pdbChains.iterator();
 				while (it.hasNext()) {
 					Chain chain = it.next();
-					GroupType predominantGroupType = StructureTools.getPredominantGroupType(chain);
 					if (StructureTools.isChainWaterOnly(chain)) {
 						it.remove();
 						logger.warn("Chain with chain id {} (asym id {}) and {} residues, contains only waters. Will ignore the chain because it doesn't fit into the BioJava structure data model.",
 								chain.getChainID(),chain.getInternalChainID(),chain.getAtomGroups().size());
 					}
-					else if (predominantGroupType != GroupType.AMINOACID && 
-							 predominantGroupType!=GroupType.NUCLEOTIDE ) {
+					else if (StructureTools.isChainPureNonPolymer(chain) ) {
 						logger.warn("Chain with chain id {} (asym id {}) and {} residues, does not seem to be polymeric. Will ignore the chain because it doesn't fit into the BioJava structure data model.",
 								chain.getChainID(),chain.getInternalChainID(),chain.getAtomGroups().size());
 						it.remove();

Original file line number	Diff line number	Diff line change
`@@ -510,6 +510,8 @@ private void initResSerialsMap(Chain c) {`
`510`	`510`	`logger.warn("No SEQRES groups found in chain {}, will use residue numbers as given (no insertion codes, not necessarily aligned). "`
`511`	`511`	`+ "Make sure your structure has SEQRES records and that you use FileParsingParameters.setAlignSeqRes(true)",`
`512`	`512`	`c.getChainID());`
	`513`	`+ // we add a explicit null to the map so that we flag it as unavailable for this chain`
	`514`	`+ chains2pdbResNums2ResSerials.put(c.getChainID(), null);`
`513`	`515`	`return;`
`514`	`516`	`}`
`515`	`517`
Original file line number	Diff line number	Diff line change
`@@ -2139,6 +2139,10 @@ private void pdb_MODEL_Handler(String line) {`
`2139`	`2139`	`logger.warn("Chain {} ({} atom groups) is composed of water molecules only. Removing it.",`
`2140`	`2140`	`c.getChainID(), c.getAtomGroups().size());`
`2141`	`2141`	`it.remove();`
	`2142`	`+ } else if (StructureTools.isChainPureNonPolymer(c)) {`
	`2143`	`+ logger.warn("Chain {} ({} atom groups) is composed of non-polymer molecules only. Removing it.",`
	`2144`	`+ c.getChainID(), c.getAtomGroups().size());`
	`2145`	`+ it.remove();`
`2142`	`2146`	`}`
`2143`	`2147`	`}`
`2144`	`2148`	`structure.addModel(current_model);`
`@@ -2922,6 +2926,10 @@ private void triggerEndFileChecks(){`
`2922`	`2926`	`logger.warn("Chain {} ({} atom groups) is composed of water molecules only. Removing it.",`
`2923`	`2927`	`c.getChainID(), c.getAtomGroups().size());`
`2924`	`2928`	`it.remove();`
	`2929`	`+ } else if (StructureTools.isChainPureNonPolymer(c)) {`
	`2930`	`+ logger.warn("Chain {} ({} atom groups) is composed of non-polymer molecules only. Removing it.",`
	`2931`	`+ c.getChainID(), c.getAtomGroups().size());`
	`2932`	`+ it.remove();`
`2925`	`2933`	`}`
`2926`	`2934`	`}`
`2927`	`2935`	`structure.addModel(current_model);`