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

fixing the build 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1705 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
chartl 2009-09-23 20:04:09 +00:00
parent 540e1b971f
commit 63f3d45ca4
1 changed files with 61 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

110
java/src/org/broadinstitute/sting/playground/gatk/walkers/Recalibration/NQSMismatchCovariantWalker.java
View File

@ -2,11 +2,9 @@ package org.broadinstitute.sting.playground.gatk.walkers.Recalibration;
import org.broadinstitute.sting.gatk.walkers.LocusWalker; import org.broadinstitute.sting.gatk.walkers.LocusWalker;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker; import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.refdata.rodDbSNP;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext; import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext; import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.utils.Pair; import org.broadinstitute.sting.utils.QualityUtils;
import org.broadinstitute.sting.utils.GenomeLoc;
import net.sf.samtools.SAMRecord; import net.sf.samtools.SAMRecord;
import java.util.List; import java.util.List;
@ -18,7 +16,7 @@ import java.util.List;
* Time: 1:37:44 PM * Time: 1:37:44 PM
* To change this template use File | Settings | File Templates. * To change this template use File | Settings | File Templates.
*/ */
public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> { public class NQSMismatchCovariantWalker extends LocusWalker<LocalMapType, long[][][]> {
public static final int NQS_GROUPS = 35; public static final int NQS_GROUPS = 35;
public static final int NQS_RESOLUTION = 1; public static final int NQS_RESOLUTION = 1;
public static final int NQS_DIFFERENCE = 5; public static final int NQS_DIFFERENCE = 5;
@ -27,47 +25,43 @@ public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> {
public static final int MM_OFFSET = 0; public static final int MM_OFFSET = 0;
public static final int COUNT_OFFSET = 1; public static final int COUNT_OFFSET = 1;
public static final int FAIL_OFFSET = 2; public static final int FAIL_OFFSET = 2;
public static final String DATA_FORMAT = "%d %d %d %d %d%n"; public static final String DATA_FORMAT = "%d %d %d %d %d %d%n";
public static final String HEADER_FORMAT = "%s %s %s %s %s%n"; public static final String HEADER_FORMAT = "%s %s %s %s %s %s%n";
public long[][] reduceInit() { public long[][][] reduceInit() {
long[][] mismatchesByNQS = new long[this.numNQSGroups()][3]; long[][][] mismatchesByNQS = new long[(int)QualityUtils.MAX_QUAL_SCORE+1][this.numNQSGroups()+1][2];
for ( int nqsGroup = 0; nqsGroup < mismatchesByNQS.length; nqsGroup++ ) { for ( int qualityScore = 0; qualityScore <= (int) QualityUtils.MAX_QUAL_SCORE; qualityScore ++) {
mismatchesByNQS[nqsGroup][MM_OFFSET] = 0; for ( int nqsGroup = 0; nqsGroup <= numNQSGroups(); nqsGroup++ ) {
mismatchesByNQS[nqsGroup][COUNT_OFFSET] = 0; mismatchesByNQS[qualityScore][nqsGroup][MM_OFFSET] = 0;
mismatchesByNQS[qualityScore][nqsGroup][COUNT_OFFSET] = 0;
}
} }
return mismatchesByNQS; return mismatchesByNQS;
} }
public int[][] map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) { public LocalMapType map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
int[][] mismatchesByNQSAtLocation = initializeMismatchMatrix();
if (! isDBSNP(tracker) ) { return new LocalMapType(context,ref,tracker);
for( int read = 0; read < context.numReads(); read++ ) { }
int bestNQSRelativeToStart = getBestNQSRelativeToStart(context, read, nqsDifference(), nqsResolution(),
centerBaseQualityStart(), windowSize()); public long[][][] reduce(LocalMapType rawMapData, long[][][] cumMismatchArray) {
if ( ! isDBSNP(rawMapData.tracker) ) {
for( int read = 0; read < rawMapData.numReads(); read ++ ) {
int bestNQSRelativeToStart = getBestNQSRelativeToStart(rawMapData.context, read, nqsDifference(),
nqsResolution(), centerBaseQualityStart(), windowSize());
if( isValidRelativeNQS(bestNQSRelativeToStart) ) { if( isValidRelativeNQS(bestNQSRelativeToStart) ) {
mismatchesByNQSAtLocation[bestNQSRelativeToStart][COUNT_OFFSET]++; cumMismatchArray[rawMapData.qScore(read)][bestNQSRelativeToStart][COUNT_OFFSET]++;
if( isMismatch(read,context,ref) ) { if( isMismatch(read, rawMapData.context, rawMapData.ref) ) {
mismatchesByNQSAtLocation[bestNQSRelativeToStart][MM_OFFSET]++; cumMismatchArray[rawMapData.qScore(read)][bestNQSRelativeToStart][MM_OFFSET]++;
} }
} }
} }
} }
return mismatchesByNQSAtLocation;
}
public long[][] reduce(int[][] mismatchArrayAtLoc, long[][] cumMismatchArray) {
for( int nqsGroup = 0; nqsGroup < this.numNQSGroups(); nqsGroup++ ) {
cumMismatchArray[nqsGroup][0] += mismatchArrayAtLoc[nqsGroup][0];
cumMismatchArray[nqsGroup][1] += mismatchArrayAtLoc[nqsGroup][1];
}
return cumMismatchArray; return cumMismatchArray;
} }
public void onTraversalDone(long[][] cumulativeCounts) { public void onTraversalDone(long[][][] cumulativeCounts) {
printNQSMismatchTable(cumulativeCounts); printNQSMismatchTable(cumulativeCounts);
} }
@ -75,8 +69,8 @@ public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> {
int ctrBaseQStart, int windSize) { int ctrBaseQStart, int windSize) {
int minNeighborhoodQualityScore = findMinQScoreInWindow(context, readNum, windSize); int minNeighborhoodQualityScore = findMinQScoreInWindow(context, readNum, windSize);
int centerQScore = getQualityScore(context,readNum); int centerQScore = getQualityScore(context,readNum);
int stepsToCutoffNeighborhood = (int) Math.floor((minNeighborhoodQualityScore - (ctrBaseQStart + 1 - ctrNghdNQSDif))/nqsStepSize); int stepsToCutoffNeighborhood = (int) Math.floor((minNeighborhoodQualityScore - (ctrBaseQStart - ctrNghdNQSDif))/nqsStepSize);
int stepsToCutoffCenter = (int) Math.floor((centerQScore-ctrBaseQStart - 1 )/nqsStepSize); int stepsToCutoffCenter = (int) Math.floor((centerQScore-ctrBaseQStart )/nqsStepSize);
if(stepsToCutoffNeighborhood < stepsToCutoffCenter) { if(stepsToCutoffNeighborhood < stepsToCutoffCenter) {
return stepsToCutoffNeighborhood; return stepsToCutoffNeighborhood;
@ -85,21 +79,23 @@ public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> {
} }
} }
protected void printNQSMismatchTable(long[][] cumulativeCounts) { protected void printNQSMismatchTable(long[][][] cumulativeCounts) {
out.printf("%s", createHeader() ); out.printf("%s", createHeader() );
for( int nqsGroup = 0; nqsGroup < numNQSGroups(); nqsGroup ++ ) { for ( int qscore = 0; qscore <= QualityUtils.MAX_QUAL_SCORE; qscore ++ ) {
out.printf("%s", formatNQSMismatchCountString(cumulativeCounts, nqsGroup)); for( int nqsGroup = 0; nqsGroup <= numNQSGroups(); nqsGroup ++ ) {
out.printf("%s", formatNQSMismatchCountString(cumulativeCounts, nqsGroup, qscore));
}
} }
} }
protected String createHeader() { protected String createHeader() {
return String.format(HEADER_FORMAT,"Qscore Threshold at Locus", "Minimum Neighborhood Quality", "NQS Group", return String.format(HEADER_FORMAT,"Qscore Reported","Qscore Threshold at Locus", "Minimum Neighborhood Quality", "NQS Group",
"# Non-dbSNP Mismatches", "Total Non-dbSNP Counts"); "# Non-dbSNP Mismatches", "Total Non-dbSNP Counts");
} }
protected String formatNQSMismatchCountString(long[][] counts, int nqsGroup) { protected String formatNQSMismatchCountString(long[][][] counts, int nqsGroup, int qscore) {
return String.format(DATA_FORMAT,getCenterThreshold(nqsGroup),getNeighborhoodThreshold(nqsGroup),nqsGroup, return String.format(DATA_FORMAT,qscore,getCenterThreshold(nqsGroup),getNeighborhoodThreshold(nqsGroup),nqsGroup,
counts[nqsGroup][MM_OFFSET], counts[nqsGroup][COUNT_OFFSET]); counts[qscore][nqsGroup][MM_OFFSET], counts[qscore][nqsGroup][COUNT_OFFSET]);
} }
protected int getCenterThreshold(int nqsGroup) { protected int getCenterThreshold(int nqsGroup) {
@ -122,26 +118,18 @@ public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> {
return NQS_DIFFERENCE; return NQS_DIFFERENCE;
} }
protected int[][] initializeMismatchMatrix() {
int[][] mismatchesByNQS = new int[this.numNQSGroups()][2];
for ( int nqsGroup = 0; nqsGroup < numNQSGroups(); nqsGroup++ ) {
mismatchesByNQS[nqsGroup][MM_OFFSET] = 0;
mismatchesByNQS[nqsGroup][COUNT_OFFSET] = 0;
}
return mismatchesByNQS;
}
protected boolean isValidRelativeNQS(int relNQS) { protected boolean isValidRelativeNQS(int relNQS) {
return relNQS >= 0; return relNQS >= 0;
} }
protected boolean isDBSNP(RefMetaDataTracker tracker) { protected boolean isDBSNP(RefMetaDataTracker tracker) {
return ( tracker.lookup("dbSNP",null) != null ); return false;
// return ( tracker.lookup("dbSNP",null) != null );
} }
protected boolean isMismatch(int readNum, AlignmentContext context, ReferenceContext ref) { protected boolean isMismatch(int readNum, AlignmentContext context, ReferenceContext ref) {
return ( (char) getBases(context,readNum)[ getOffset(context,readNum) ] ) == ref.getBase(); return ( (char) getBases(context,readNum)[ getOffset(context,readNum) ] ) != ref.getBase();
} }
protected byte[] getBases(AlignmentContext context, int offset) { protected byte[] getBases(AlignmentContext context, int offset) {
@ -198,3 +186,25 @@ public class NQSMismatchCovariantWalker extends LocusWalker<int[][], long[][]> {
} }
} }
class LocalMapType {
public AlignmentContext context;
public ReferenceContext ref;
public RefMetaDataTracker tracker;
public LocalMapType(AlignmentContext context, ReferenceContext ref, RefMetaDataTracker tracker) {
this.context = context;
this.ref = ref;
this.tracker = tracker;
}
public int numReads() {
return context.numReads();
}
public int qScore(int read) {
return (int) context.getReads().get(read).getBaseQualities()[context.getOffsets().get(read)];
}
}