Merge branch 'master' of ssh://nickel.broadinstitute.org/humgen/gsa-scr1/gsa-engineering/git/unstable

This commit is contained in:
Guillermo del Angel 2011-09-28 18:17:34 -04:00
commit c8d3a720f9
6 changed files with 82 additions and 10 deletions

View File

@ -159,10 +159,11 @@ public class CycleCovariate implements StandardCovariate {
*/
// todo -- this should be put into a common place in the code base
private static List<String> PACBIO_NAMES = Arrays.asList("PACBIO");
private static List<String> ILLUMINA_NAMES = Arrays.asList("ILLUMINA", "SLX", "SOLEXA");
private static List<String> SOLID_NAMES = Arrays.asList("SOLID");
private static List<String> LS454_NAMES = Arrays.asList("454");
private static List<String> COMPLETE_GENOMICS_NAMES = Arrays.asList("COMPLETE");
private static List<String> PACBIO_NAMES = Arrays.asList("PACBIO");
private static boolean isPlatform(SAMRecord read, List<String> names) {
String pl = read.getReadGroup().getPlatform().toUpperCase();
@ -176,11 +177,10 @@ public class CycleCovariate implements StandardCovariate {
public void getValues(SAMRecord read, Comparable[] comparable) {
//-----------------------------
// ILLUMINA and SOLID
// Illumina, Solid, PacBio, and Complete Genomics
//-----------------------------
if( isPlatform(read, ILLUMINA_NAMES) || isPlatform(read, SOLID_NAMES) || isPlatform(read, PACBIO_NAMES)) {
if( isPlatform(read, ILLUMINA_NAMES) || isPlatform(read, SOLID_NAMES) || isPlatform(read, PACBIO_NAMES) || isPlatform(read, COMPLETE_GENOMICS_NAMES) ) {
final int init;
final int increment;
if( !read.getReadNegativeStrandFlag() ) {
@ -222,6 +222,11 @@ public class CycleCovariate implements StandardCovariate {
cycle += increment;
}
}
//-----------------------------
// 454
//-----------------------------
else if ( isPlatform(read, LS454_NAMES) ) { // Some bams have "LS454" and others have just "454"
final int readLength = read.getReadLength();

View File

@ -245,8 +245,7 @@ public class RecalDataManager {
readGroup.setPlatform( RAC.DEFAULT_PLATFORM );
((GATKSAMRecord)read).setReadGroup( readGroup );
} else {
throw new UserException.MalformedBAM(read, "The input .bam file contains reads with no read group. First observed at read with name = " + read.getReadName() +
" Users must set both the default read group using the --default_read_group <String> argument and the default platform using the --default_platform <String> argument." );
throw new UserException.MalformedBAM(read, "The input .bam file contains reads with no read group. First observed at read with name = " + read.getReadName() );
}
}
@ -271,8 +270,7 @@ public class RecalDataManager {
}
readGroup.setPlatform( RAC.DEFAULT_PLATFORM );
} else {
throw new UserException.MalformedBAM(read, "The input .bam file contains reads with no platform information. First observed at read with name = " + read.getReadName() +
" Users must set the default platform using the --default_platform <String> argument." );
throw new UserException.MalformedBAM(read, "The input .bam file contains reads with no platform information. First observed at read with name = " + read.getReadName() );
}
}
}

View File

@ -26,6 +26,7 @@
package org.broadinstitute.sting.gatk.walkers.recalibration;
import org.broadinstitute.sting.commandline.Argument;
import org.broadinstitute.sting.commandline.Hidden;
/**
* Created by IntelliJ IDEA.
@ -41,22 +42,29 @@ public class RecalibrationArgumentCollection {
//////////////////////////////////
// Shared Command Line Arguments
//////////////////////////////////
@Hidden
@Argument(fullName="default_read_group", shortName="dRG", required=false, doc="If a read has no read group then default to the provided String.")
public String DEFAULT_READ_GROUP = null;
@Hidden
@Argument(fullName="default_platform", shortName="dP", required=false, doc="If a read has no platform then default to the provided String. Valid options are illumina, 454, and solid.")
public String DEFAULT_PLATFORM = null;
@Hidden
@Argument(fullName="force_read_group", shortName="fRG", required=false, doc="If provided, the read group ID of EVERY read will be forced to be the provided String. This is useful to collapse all data into a single read group.")
public String FORCE_READ_GROUP = null;
@Hidden
@Argument(fullName="force_platform", shortName="fP", required=false, doc="If provided, the platform of EVERY read will be forced to be the provided String. Valid options are illumina, 454, and solid.")
public String FORCE_PLATFORM = null;
@Hidden
@Argument(fullName = "window_size_nqs", shortName="nqs", doc="The window size used by MinimumNQSCovariate for its calculation", required=false)
public int WINDOW_SIZE = 5;
/**
* This window size tells the module in how big of a neighborhood around the current base it should look for the minimum base quality score.
*/
@Hidden
@Argument(fullName = "homopolymer_nback", shortName="nback", doc="The number of previous bases to look at in HomopolymerCovariate", required=false)
public int HOMOPOLYMER_NBACK = 7;
@Hidden
@Argument(fullName = "exception_if_no_tile", shortName="throwTileException", doc="If provided, TileCovariate will throw an exception when no tile can be found. The default behavior is to use tile = -1", required=false)
public boolean EXCEPTION_IF_NO_TILE = false;

View File

@ -0,0 +1,56 @@
package org.broadinstitute.sting.utils;
import com.google.java.contract.Ensures;
import com.google.java.contract.Requires;
import java.util.Comparator;
/**
*
* @author Mauricio Carneiro
* @since 9/28/11
*/
public class GenomeLocComparator implements Comparator<GenomeLoc> {
/**
* compares genomeLoc's contigs
*
* @param gl1 the genome loc to compare contigs
* @param gl2 the genome loc to compare contigs
* @return 0 if equal, -1 if gl2.contig is greater, 1 if gl1.contig is greater
*/
@Requires("gl2 != null")
@Ensures("result == 0 || result == 1 || result == -1")
public final int compareContigs( GenomeLoc gl1, GenomeLoc gl2 ) {
if (gl1.contigIndex == gl2.contigIndex)
return 0;
else if (gl1.contigIndex > gl2.contigIndex)
return 1;
return -1;
}
@Requires("gl2 != null")
@Ensures("result == 0 || result == 1 || result == -1")
public int compare ( GenomeLoc gl1, GenomeLoc gl2 ) {
int result = 0;
if ( gl1 == gl2 ) {
result = 0;
}
else if(GenomeLoc.isUnmapped(gl1))
result = 1;
else if(GenomeLoc.isUnmapped(gl2))
result = -1;
else {
final int cmpContig = compareContigs(gl1, gl2);
if ( cmpContig != 0 ) {
result = cmpContig;
} else {
if ( gl1.getStart() < gl2.getStart() ) result = -1;
if ( gl1.getStart() > gl2.getStart() ) result = 1;
}
}
return result;
}
}

View File

@ -93,8 +93,9 @@ public class ReadClipper {
public SAMRecord hardClipBothEndsByReferenceCoordinates(int left, int right) {
if (left == right)
return new SAMRecord(read.getHeader());
this.read = hardClipByReferenceCoordinates(right, -1);
return hardClipByReferenceCoordinates(-1, left);
SAMRecord leftTailRead = hardClipByReferenceCoordinates(right, -1);
ReadClipper clipper = new ReadClipper(leftTailRead);
return clipper.hardClipByReferenceCoordinatesLeftTail(left);
}
public SAMRecord hardClipLowQualEnds(byte lowQual) {

View File

@ -893,6 +893,10 @@ public class ReadUtils {
// base before the deletion (see warning in function contracts)
else if (fallsInsideDeletion && !endsWithinCigar)
readBases += shift - 1;
// If we reached our goal inside a deletion then we must backtrack to the last base before the deletion
else if (fallsInsideDeletion && endsWithinCigar)
readBases--;
}
}