Allow single-residue ranges

e.g. "A:1" rather than "A:1-1"
biojava · Apr 15, 2015 · 8be282f · 8be282f
1 parent a4bdcb7
commit 8be282f
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 7 deletions.
diff --git a/...ava-integrationtest/src/test/java/org/biojava/nbio/structure/test/StructureToolsTest.java b/...ava-integrationtest/src/test/java/org/biojava/nbio/structure/test/StructureToolsTest.java
@@ -164,6 +164,13 @@ public void testGetSubRanges() throws StructureException {
 		chain = substr.getChain(0);
 		assertEquals("Did not find the expected number of residues in "+range, 5, chain.getAtomLength() );
 
+		// single residue
+		range = "A:3";
+		substr = StructureTools.getSubRanges(structure2, range);
+		assertEquals("Wrong number of chains in "+range, 1, substr.size());
+		chain = substr.getChain(0);
+		assertEquals("Did not find the expected number of residues in "+range, 1, chain.getAtomLength() );
+
 		// Special '-' case
 		range = "-";
 		substr = StructureTools.getSubRanges(structure2, range);

diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/ResidueRange.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/ResidueRange.java
@@ -47,8 +47,10 @@ public class ResidueRange {
 				"(?::|_|:$|_$|$)" + //colon or underscore, could be at the end of a line, another non-capt. group.
 				"(?:"+ // another non capturing group for the residue range
 					"([-+]?[0-9]+[A-Za-z]?)" + // first residue
-					"\\s*-\\s*" + // -
-					"([-+]?[0-9]+[A-Za-z]?)" + // second residue
+					"(?:" +
+						"\\s*-\\s*" + // -
+						"([-+]?[0-9]+[A-Za-z]?)" + // second residue
+					")?+"+
 				")?+"+
 			")?" + //end range
 			"\\s*");

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
@@ -1151,6 +1151,11 @@ public static final Structure getSubRanges(Structure s, String ranges )
 
 			String pdbresnumStart = matcher.group(2);
 			String pdbresnumEnd   = matcher.group(3);
+
+			if(pdbresnumEnd == null ) {
+				// Single residue range
+				pdbresnumEnd = pdbresnumStart;
+			}
 
 
 			if ( ! firstRange){

diff --git a/biojava-structure/src/test/java/org/biojava/nbio/structure/ResidueRangeTest.java b/biojava-structure/src/test/java/org/biojava/nbio/structure/ResidueRangeTest.java
@@ -327,10 +327,8 @@ public void testRangeRegex() {
 		// Valid ranges
 		String[] yes = new String[] { "A_", "A:", "ABC:", "abc:", "A_5-100",
 				"A_5-100S", "A_5S-100", "A_5S-100S", "A_-5-100", "A_-5--100",
-				"A_-5S--100S", "ABC:-5--200S", "A", "ABCD", "A1", // valid
-				// multi-char
-				// chain
-				// name
+				"A_-5S--100S", "ABC:-5--200S", "A", "ABCD", "A_1", 
+				"A1", // valid multi-char chain name
 				"3A:1-100", // Weird chain name
 				"_", "_:1-10", "__-2--1", "__"// catch-all chain
 		};
@@ -339,7 +337,7 @@ public void testRangeRegex() {
 					ResidueRange.RANGE_REGEX.matcher(s).matches());
 		}
 		// invalid ranges
-		String[] no = new String[] { "A_1", "A_1-", "A_1S-", "A_1-100-",
+		String[] no = new String[] { "A_1-", "A_1S-", "A_1-100-",
 				"A_-10-1000_", "", "-", "___", "__:" };
 		for (String s : no) {
 			assertFalse(s + " was considered a valid range format",