zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

getAdaptorBoundary returns an int, not an Integer, as this was taking 30% of the allocation effort for LIBS

This commit is contained in:
Mark DePristo 2013-01-11 18:05:45 -05:00
parent 682c59ff04
commit ec05ecef60
3 changed files with 21 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
public/java/src/org/broadinstitute/sting/utils/clipping/ReadClipper.java
View File

@ -381,9 +381,9 @@ public class ReadClipper {
* @return a new read without adaptor sequence
*/
private GATKSAMRecord hardClipAdaptorSequence () {
final Integer adaptorBoundary = ReadUtils.getAdaptorBoundary(read);
final int adaptorBoundary = ReadUtils.getAdaptorBoundary(read);
if (adaptorBoundary == null || !ReadUtils.isInsideRead(read, adaptorBoundary))
if (adaptorBoundary == ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY || !ReadUtils.isInsideRead(read, adaptorBoundary))
return read;
return read.getReadNegativeStrandFlag() ? hardClipByReferenceCoordinatesLeftTail(adaptorBoundary) : hardClipByReferenceCoordinatesRightTail(adaptorBoundary);

16
public/java/src/org/broadinstitute/sting/utils/sam/ReadUtils.java
View File

@ -169,8 +169,8 @@ public class ReadUtils {
* @return whether or not the base is in the adaptor
*/
public static boolean isBaseInsideAdaptor(final GATKSAMRecord read, long basePos) {
Integer adaptorBoundary = getAdaptorBoundary(read);
if (adaptorBoundary == null || read.getInferredInsertSize() > DEFAULT_ADAPTOR_SIZE)
final int adaptorBoundary = getAdaptorBoundary(read);
if (adaptorBoundary == CANNOT_COMPUTE_ADAPTOR_BOUNDARY || read.getInferredInsertSize() > DEFAULT_ADAPTOR_SIZE)
return false;
return read.getReadNegativeStrandFlag() ? basePos <= adaptorBoundary : basePos >= adaptorBoundary;
@ -199,26 +199,28 @@ public class ReadUtils {
* in these cases the adaptor boundary is at the start of the read plus the inferred insert size (plus one)
*
* @param read the read being tested for the adaptor boundary
* @return the reference coordinate for the adaptor boundary (effectively the first base IN the adaptor, closest to the read. NULL if the read is unmapped or the mate is mapped to another contig.
* @return the reference coordinate for the adaptor boundary (effectively the first base IN the adaptor, closest to the read.
* CANNOT_COMPUTE_ADAPTOR_BOUNDARY if the read is unmapped or the mate is mapped to another contig.
*/
public static Integer getAdaptorBoundary(final SAMRecord read) {
public static int getAdaptorBoundary(final SAMRecord read) {
final int MAXIMUM_ADAPTOR_LENGTH = 8;
final int insertSize = Math.abs(read.getInferredInsertSize()); // the inferred insert size can be negative if the mate is mapped before the read (so we take the absolute value)
if (insertSize == 0 || read.getReadUnmappedFlag()) // no adaptors in reads with mates in another chromosome or unmapped pairs
return null;
return CANNOT_COMPUTE_ADAPTOR_BOUNDARY;
Integer adaptorBoundary; // the reference coordinate for the adaptor boundary (effectively the first base IN the adaptor, closest to the read)
int adaptorBoundary; // the reference coordinate for the adaptor boundary (effectively the first base IN the adaptor, closest to the read)
if (read.getReadNegativeStrandFlag())
adaptorBoundary = read.getMateAlignmentStart() - 1; // case 1 (see header)
else
adaptorBoundary = read.getAlignmentStart() + insertSize + 1; // case 2 (see header)
if ( (adaptorBoundary < read.getAlignmentStart() - MAXIMUM_ADAPTOR_LENGTH) || (adaptorBoundary > read.getAlignmentEnd() + MAXIMUM_ADAPTOR_LENGTH) )
adaptorBoundary = null; // we are being conservative by not allowing the adaptor boundary to go beyond what we belive is the maximum size of an adaptor
adaptorBoundary = CANNOT_COMPUTE_ADAPTOR_BOUNDARY; // we are being conservative by not allowing the adaptor boundary to go beyond what we belive is the maximum size of an adaptor
return adaptorBoundary;
}
public static int CANNOT_COMPUTE_ADAPTOR_BOUNDARY = Integer.MIN_VALUE;
/**
* is the read a 454 read?

20
public/java/test/org/broadinstitute/sting/utils/sam/ReadUtilsUnitTest.java
View File

@ -40,7 +40,7 @@ public class ReadUtilsUnitTest extends BaseTest {
final int mateStart = 1000;
final int BEFORE = mateStart - 2;
final int AFTER = mateStart + 2;
Integer myStart, boundary;
int myStart, boundary;
GATKSAMRecord read = ArtificialSAMUtils.createArtificialRead(bases, quals, cigar);
read.setMateAlignmentStart(mateStart);
@ -51,43 +51,43 @@ public class ReadUtilsUnitTest extends BaseTest {
read.setAlignmentStart(myStart);
read.setReadNegativeStrandFlag(false);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertEquals(boundary.intValue(), myStart + fragmentSize + 1);
Assert.assertEquals(boundary, myStart + fragmentSize + 1);
// Test case 2: positive strand, second read
myStart = AFTER;
read.setAlignmentStart(myStart);
read.setReadNegativeStrandFlag(false);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertEquals(boundary.intValue(), myStart + fragmentSize + 1);
Assert.assertEquals(boundary, myStart + fragmentSize + 1);
// Test case 3: negative strand, second read
myStart = AFTER;
read.setAlignmentStart(myStart);
read.setReadNegativeStrandFlag(true);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertEquals(boundary.intValue(), mateStart - 1);
Assert.assertEquals(boundary, mateStart - 1);
// Test case 4: negative strand, first read
myStart = BEFORE;
read.setAlignmentStart(myStart);
read.setReadNegativeStrandFlag(true);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertEquals(boundary.intValue(), mateStart - 1);
Assert.assertEquals(boundary, mateStart - 1);
// Test case 5: mate is mapped to another chromosome (test both strands)
read.setInferredInsertSize(0);
read.setReadNegativeStrandFlag(true);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertNull(boundary);
Assert.assertEquals(boundary, ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY);
read.setReadNegativeStrandFlag(false);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertNull(boundary);
Assert.assertEquals(boundary, ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY);
read.setInferredInsertSize(10);
// Test case 6: read is unmapped
read.setReadUnmappedFlag(true);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertNull(boundary);
Assert.assertEquals(boundary, ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY);
read.setReadUnmappedFlag(false);
// Test case 7: reads don't overlap and look like this:
@ -99,7 +99,7 @@ public class ReadUtilsUnitTest extends BaseTest {
read.setInferredInsertSize(20);
read.setReadNegativeStrandFlag(true);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertNull(boundary);
Assert.assertEquals(boundary, ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY);
// second read:
myStart = 1000;
@ -107,6 +107,6 @@ public class ReadUtilsUnitTest extends BaseTest {
read.setMateAlignmentStart(980);
read.setReadNegativeStrandFlag(false);
boundary = ReadUtils.getAdaptorBoundary(read);
Assert.assertNull(boundary);
Assert.assertEquals(boundary, ReadUtils.CANNOT_COMPUTE_ADAPTOR_BOUNDARY);
}
}