Start of refactoring effort. Make construction of alignment object simpler.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1747 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
hanna 2009-09-30 15:19:31 +00:00
parent 70e1aef550
commit d3b1732cca
2 changed files with 120 additions and 93 deletions

View File

@ -6,10 +6,7 @@ import org.broadinstitute.sting.alignment.Aligner;
import org.broadinstitute.sting.utils.BaseUtils; import org.broadinstitute.sting.utils.BaseUtils;
import java.io.File; import java.io.File;
import java.util.List; import java.util.*;
import java.util.PriorityQueue;
import java.util.Collections;
import java.util.EnumSet;
import net.sf.samtools.SAMRecord; import net.sf.samtools.SAMRecord;
@ -76,9 +73,10 @@ public class BWAAligner implements Aligner {
byte[] reverseBases = BaseUtils.simpleReverseComplement(forwardBases); byte[] reverseBases = BaseUtils.simpleReverseComplement(forwardBases);
List<LowerBound> forwardLowerBounds = LowerBound.create(forwardBases,forwardBWT); List<LowerBound> forwardLowerBounds = LowerBound.create(forwardBases,forwardBWT);
List<LowerBound> reverseLowerBounds = LowerBound.create(reverseBases,reverseBWT);
//for( int i = 0; i < forwardLowerBounds.size(); i++ ) //for( int i = 0; i < forwardLowerBounds.size(); i++ )
// System.out.printf("ForwardBWT: lb[%d] = %s%n",i,forwardLowerBounds.get(i)); // System.out.printf("ForwardBWT: lb[%d] = %s%n",i,forwardLowerBounds.get(i));
List<LowerBound> reverseLowerBounds = LowerBound.create(reverseBases,reverseBWT);
//for( int i = 0; i < reverseLowerBounds.size(); i++ ) //for( int i = 0; i < reverseLowerBounds.size(); i++ )
// System.out.printf("ReverseBWT: lb[%d] = %s%n",i,reverseLowerBounds.get(i)); // System.out.printf("ReverseBWT: lb[%d] = %s%n",i,reverseLowerBounds.get(i));
@ -86,24 +84,8 @@ public class BWAAligner implements Aligner {
// Create a fictional initial alignment, with the position just off the end of the read, and the limits // Create a fictional initial alignment, with the position just off the end of the read, and the limits
// set as the entire BWT. // set as the entire BWT.
BWAAlignment initial = new BWAAlignment(this); alignments.add(createSeedAlignment(forwardBWT));
initial.negativeStrand = true; //alignments.add(createSeedAlignment(reverseBWT));
initial.position = 0;
initial.loBound = 0;
initial.hiBound = forwardBWT.length();
initial.state = AlignmentState.MATCH_MISMATCH;
initial.mismatches = 0;
BWAAlignment initialReverse = new BWAAlignment(this);
initialReverse.negativeStrand = false;
initialReverse.position = 0;
initialReverse.loBound = 0;
initialReverse.hiBound = reverseBWT.length();
initialReverse.state = AlignmentState.MATCH_MISMATCH;
initialReverse.mismatches = 0;
alignments.add(initial);
//alignments.add(initialReverse);
while(!alignments.isEmpty()) { while(!alignments.isEmpty()) {
BWAAlignment alignment = alignments.remove(); BWAAlignment alignment = alignments.remove();
@ -131,94 +113,125 @@ public class BWAAligner implements Aligner {
if( alignment.state == AlignmentState.MATCH_MISMATCH ) { if( alignment.state == AlignmentState.MATCH_MISMATCH ) {
if( alignment.gapOpens < MAXIMUM_GAP_OPENS ) { if( alignment.gapOpens < MAXIMUM_GAP_OPENS ) {
// Add a potential insertion. // Add a potential insertion.
BWAAlignment newAlignment = new BWAAlignment(this); // Add a potential insertion extension.
newAlignment.negativeStrand = alignment.negativeStrand; BWAAlignment insertionAlignment = createInsertionAlignment(alignment);
newAlignment.position = alignment.position + 1; insertionAlignment.gapOpens++;
newAlignment.state = AlignmentState.INSERTION; alignments.add(insertionAlignment);
newAlignment.gapOpens = alignment.gapOpens + 1;
newAlignment.gapExtensions = alignment.gapExtensions;
newAlignment.loBound = alignment.loBound;
newAlignment.hiBound = alignment.hiBound;
alignments.add(newAlignment);
// Add a potential deletion by marking a deletion and augmenting the position. // Add a potential deletion by marking a deletion and augmenting the position.
for(Base base: EnumSet.allOf(Base.class)) { List<BWAAlignment> newAlignments = createDeletionAlignments(bwt,alignment);
newAlignment = new BWAAlignment(this); for( BWAAlignment deletionAlignment: newAlignments )
newAlignment.negativeStrand = alignment.negativeStrand; deletionAlignment.gapOpens++;
newAlignment.position = alignment.position; alignments.addAll(newAlignments);
newAlignment.state = AlignmentState.DELETION;
newAlignment.gapOpens = alignment.gapOpens + 1;
newAlignment.gapExtensions = alignment.gapExtensions;
newAlignment.loBound = bwt.counts(base) + bwt.occurrences(base,alignment.loBound-1) + 1;
newAlignment.hiBound = bwt.counts(base) + bwt.occurrences(base,alignment.hiBound);
if( newAlignment.loBound <= newAlignment.hiBound )
alignments.add(newAlignment);
}
} }
} }
else if( alignment.state == AlignmentState.INSERTION ) { else if( alignment.state == AlignmentState.INSERTION ) {
if( alignment.gapExtensions < MAXIMUM_GAP_EXTENSIONS ) { if( alignment.gapExtensions < MAXIMUM_GAP_EXTENSIONS ) {
// Add a potential insertion extension. // Add a potential insertion extension.
BWAAlignment newAlignment = new BWAAlignment(this); BWAAlignment newAlignment = createInsertionAlignment(alignment);
newAlignment.negativeStrand = alignment.negativeStrand; newAlignment.gapExtensions++;
newAlignment.position = alignment.position + 1; alignments.add(newAlignment);
newAlignment.state = AlignmentState.INSERTION;
newAlignment.gapOpens = alignment.gapOpens;
newAlignment.gapExtensions = alignment.gapExtensions+1;
newAlignment.loBound = alignment.loBound;
newAlignment.hiBound = alignment.hiBound;
if( newAlignment.loBound <= newAlignment.hiBound )
alignments.add(newAlignment);
} }
} }
else if( alignment.state == AlignmentState.DELETION ) { else if( alignment.state == AlignmentState.DELETION ) {
if( alignment.gapExtensions < MAXIMUM_GAP_EXTENSIONS ) { if( alignment.gapExtensions < MAXIMUM_GAP_EXTENSIONS ) {
// Add a potential deletion by marking a deletion and augmenting the position. // Add a potential deletion by marking a deletion and augmenting the position.
for(Base base: EnumSet.allOf(Base.class)) { List<BWAAlignment> newAlignments = createDeletionAlignments(bwt,alignment);
BWAAlignment newAlignment = new BWAAlignment(this); for( BWAAlignment newAlignment: newAlignments )
newAlignment.negativeStrand = alignment.negativeStrand; newAlignment.gapExtensions++;
newAlignment.position = alignment.position; alignments.addAll(newAlignments);
newAlignment.state = AlignmentState.DELETION;
newAlignment.gapOpens = alignment.gapOpens;
newAlignment.gapExtensions = alignment.gapExtensions + 1;
newAlignment.loBound = bwt.counts(base) + bwt.occurrences(base,alignment.loBound-1) + 1;
newAlignment.hiBound = bwt.counts(base) + bwt.occurrences(base,alignment.hiBound);
alignments.add(newAlignment);
}
} }
} }
// Mismatches // Mismatches
for(Base base: EnumSet.allOf(Base.class)) { alignments.addAll(createMatchedAlignments(bwt,alignment,bases));
// Create and initialize a new alignment, given that base as the candidate.
BWAAlignment newAlignment = new BWAAlignment(this);
newAlignment.negativeStrand = alignment.negativeStrand;
newAlignment.position = alignment.position + 1;
newAlignment.state = AlignmentState.MATCH_MISMATCH;
newAlignment.mismatches = alignment.mismatches;
if( base.toASCII() != bases[newAlignment.position] )
newAlignment.mismatches++;
newAlignment.gapOpens = alignment.gapOpens;
newAlignment.gapExtensions = alignment.gapExtensions;
newAlignment.loBound = bwt.counts(base) + bwt.occurrences(base,alignment.loBound-1) + 1;
newAlignment.hiBound = bwt.counts(base) + bwt.occurrences(base,alignment.hiBound);
// If this alignment is valid, add it to the list.
if( newAlignment.loBound <= newAlignment.hiBound )
alignments.add(newAlignment);
}
} }
return Collections.emptyList(); return Collections.emptyList();
} }
/**
* Create an seeding alignment to use as a starting point when traversing.
* @param bwt source BWT.
* @return Seed alignment.
*/
private BWAAlignment createSeedAlignment(BWT bwt) {
BWAAlignment seed = new BWAAlignment(this);
seed.negativeStrand = (bwt == forwardBWT);
seed.position = 0;
seed.loBound = 0;
seed.hiBound = bwt.length();
seed.state = AlignmentState.MATCH_MISMATCH;
seed.mismatches = 0;
return seed;
}
/**
* Creates a new alignments representing direct matches / mismatches.
* @param bwt Source BWT with which to work.
* @param alignment Alignment for the previous position.
* @param bases The bases in the read.
* @return New alignment representing this position if valid; null otherwise.
*/
private List<BWAAlignment> createMatchedAlignments( BWT bwt, BWAAlignment alignment, byte[] bases ) {
List<BWAAlignment> newAlignments = new ArrayList<BWAAlignment>();
for(Base base: EnumSet.allOf(Base.class)) {
BWAAlignment newAlignment = alignment.clone();
newAlignment.loBound = bwt.counts(base) + bwt.occurrences(base,alignment.loBound-1) + 1;
newAlignment.hiBound = bwt.counts(base) + bwt.occurrences(base,alignment.hiBound);
// If this alignment is valid, skip it.
if( newAlignment.loBound > newAlignment.hiBound )
continue;
newAlignment.position++;
newAlignment.state = AlignmentState.MATCH_MISMATCH;
if( base.toASCII() != bases[newAlignment.position] )
newAlignment.mismatches++;
newAlignments.add(newAlignment);
}
return newAlignments;
}
/**
* Create a new alignment representing an insertion at this point in the read.
* @param alignment Alignment from which to derive the insertion.
* @return New alignment reflecting the insertion.
*/
private BWAAlignment createInsertionAlignment( BWAAlignment alignment ) {
// Add a potential insertion extension.
BWAAlignment newAlignment = alignment.clone();
newAlignment.position++;
newAlignment.state = AlignmentState.INSERTION;
return newAlignment;
}
/**
* Create new alignments representing a deletion a this point in the read.
* @param alignment Alignment from which to derive the deletion.
* @return New alignments reflecting all possible deletions.
*/
private List<BWAAlignment> createDeletionAlignments( BWT bwt, BWAAlignment alignment) {
List<BWAAlignment> newAlignments = new ArrayList<BWAAlignment>();
for(Base base: EnumSet.allOf(Base.class)) {
BWAAlignment newAlignment = alignment.clone();
newAlignment.loBound = bwt.counts(base) + bwt.occurrences(base,alignment.loBound-1) + 1;
newAlignment.hiBound = bwt.counts(base) + bwt.occurrences(base,alignment.hiBound);
// If this alignment is valid, skip it.
if( newAlignment.loBound > newAlignment.hiBound )
continue;
newAlignment.state = AlignmentState.DELETION;
newAlignments.add(newAlignment);
}
return newAlignments;
}
} }

View File

@ -1,6 +1,7 @@
package org.broadinstitute.sting.alignment.bwa; package org.broadinstitute.sting.alignment.bwa;
import org.broadinstitute.sting.alignment.Alignment; import org.broadinstitute.sting.alignment.Alignment;
import org.broadinstitute.sting.utils.StingException;
/** /**
* An alignment object to be used incrementally as the BWA aligner * An alignment object to be used incrementally as the BWA aligner
@ -9,7 +10,7 @@ import org.broadinstitute.sting.alignment.Alignment;
* @author mhanna * @author mhanna
* @version 0.1 * @version 0.1
*/ */
public class BWAAlignment implements Alignment { public class BWAAlignment implements Alignment, Cloneable {
/** /**
* The aligner performing the alignments. * The aligner performing the alignments.
*/ */
@ -96,6 +97,19 @@ public class BWAAlignment implements Alignment {
this.aligner = aligner; this.aligner = aligner;
} }
/**
* Clone the alignment.
* @return New instance of the alignment.
*/
public BWAAlignment clone() {
try {
return (BWAAlignment)super.clone();
}
catch( CloneNotSupportedException ex ) {
throw new StingException("Unable to clone BWAAlignment.");
}
}
/** /**
* Compare this alignment to another alignment. * Compare this alignment to another alignment.
* @param other Other alignment to which to compare. * @param other Other alignment to which to compare.