Removing deprecated code and walkers for which I had the green light from repository.
Moved piecemealannotator and secondarybases to archive. git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2195 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
ebanks
parent
2c16c18a04
commit
084337087e
|
|
@ -1,46 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers;
|
||||
|
||||
import net.sf.samtools.SAMRecord;
|
||||
import org.broadinstitute.sting.gatk.walkers.ReadWalker;
|
||||
import org.broadinstitute.sting.gatk.walkers.WalkerName;
|
||||
|
||||
@WalkerName("Base_Quality_Dump")
|
||||
public class BaseQualityDumpWalker extends ReadWalker<Integer, Integer> {
|
||||
|
||||
protected final int MIN_TARGET_EDIT_DISTANCE = 0; //5;
|
||||
protected final int MAX_TARGET_EDIT_DISTANCE = 4; //10;
|
||||
|
||||
// Do we actually want to operate on the context?
|
||||
public boolean filter(char[] ref, SAMRecord read) {
|
||||
// we only want aligned reads
|
||||
return !read.getReadUnmappedFlag();
|
||||
}
|
||||
|
||||
public Integer map(char[] ref, SAMRecord read) {
|
||||
|
||||
int editDist = Integer.parseInt(read.getAttribute("NM").toString());
|
||||
|
||||
// ignore alignments with indels for now
|
||||
if ( read.getAlignmentBlocks().size() == 1 &&
|
||||
editDist >= MIN_TARGET_EDIT_DISTANCE &&
|
||||
editDist <= MAX_TARGET_EDIT_DISTANCE ) {
|
||||
|
||||
String qualStr = read.getBaseQualityString();
|
||||
int[] scores = new int[qualStr.length()];
|
||||
boolean reverseFlag = read.getReadNegativeStrandFlag();
|
||||
for ( int i = 0; i < qualStr.length(); i++)
|
||||
scores[(reverseFlag ? (qualStr.length()-1-i) : i)] += (int)qualStr.charAt(i) - 33;
|
||||
for ( int i = 0; i < scores.length; i++ )
|
||||
out.print(scores[i] + " ");
|
||||
out.println("");
|
||||
}
|
||||
|
||||
return 1;
|
||||
}
|
||||
|
||||
public Integer reduceInit() { return 0; }
|
||||
|
||||
public Integer reduce(Integer value, Integer sum) {
|
||||
return value + sum;
|
||||
}
|
||||
}
|
||||
|
|
@ -1,60 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers;
|
||||
|
||||
import net.sf.samtools.SAMRecord;
|
||||
import org.broadinstitute.sting.gatk.walkers.ReadWalker;
|
||||
import org.broadinstitute.sting.gatk.walkers.WalkerName;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: mdepristo
|
||||
* Date: Feb 22, 2009
|
||||
* Time: 3:22:14 PM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
@WalkerName("Base_Quality_Histogram")
|
||||
public class BaseQualityHistoWalker extends ReadWalker<Integer, Integer> {
|
||||
long[] qualCounts = new long[100];
|
||||
|
||||
public void initialize() {
|
||||
for ( int i = 0; i < this.qualCounts.length; i++ ) {
|
||||
this.qualCounts[i] = 0;
|
||||
}
|
||||
}
|
||||
|
||||
// Do we actually want to operate on the context?
|
||||
public boolean filter(char[] ref, SAMRecord read) {
|
||||
return true; // We are keeping all the reads
|
||||
}
|
||||
|
||||
public Integer map(char[] ref, SAMRecord read) {
|
||||
for ( byte qual : read.getBaseQualities() ) {
|
||||
if ( qual < 0 || qual > 100 ) {
|
||||
throw new RuntimeException(String.format("Invalid base quality detected -- %d at %s%n", qual, read.getReadName()));
|
||||
}
|
||||
//System.out.println(qual);
|
||||
this.qualCounts[qual]++;
|
||||
}
|
||||
//System.out.println(read.getReadName());
|
||||
return 1;
|
||||
}
|
||||
|
||||
// Given result of map function
|
||||
public Integer reduceInit() { return 0; }
|
||||
public Integer reduce(Integer value, Integer sum) {
|
||||
return value + sum;
|
||||
}
|
||||
|
||||
public void onTraversalDone(Integer result) {
|
||||
int lastNonZero = -1;
|
||||
for ( int i = this.qualCounts.length-1; i >= 0; i-- ) {
|
||||
if ( this.qualCounts[i] > 0 ) {
|
||||
lastNonZero = i;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
for ( int i = 0; i < lastNonZero+1; i++ ) {
|
||||
out.printf("%3d : %10d%n", i, this.qualCounts[i]);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -1,57 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers;
|
||||
|
||||
import org.broadinstitute.sting.gatk.walkers.ReadWalker;
|
||||
import org.broadinstitute.sting.utils.QualityUtils;
|
||||
import net.sf.samtools.SAMRecord;
|
||||
|
||||
public class DisplayFourBaseReadWalker extends ReadWalker<Integer, Integer> {
|
||||
public Integer map(char[] ref, SAMRecord read) {
|
||||
String bases = read.getReadString();
|
||||
|
||||
boolean displayed = false;
|
||||
|
||||
byte[] sq = (byte[]) read.getAttribute("SQ");
|
||||
|
||||
if (read.getReadName().equalsIgnoreCase("30JJE.5.24197751")) {
|
||||
System.out.println(read.format());
|
||||
}
|
||||
|
||||
for (int i = 0; i < sq.length; i++) {
|
||||
int baseIndex = QualityUtils.compressedQualityToBaseIndex(sq[i]);
|
||||
char base = '.';
|
||||
switch (baseIndex) {
|
||||
case 0: base = 'A'; break;
|
||||
case 1: base = 'C'; break;
|
||||
case 2: base = 'G'; break;
|
||||
case 3: base = 'T'; break;
|
||||
}
|
||||
|
||||
if (base == bases.charAt(i)) {
|
||||
if (!displayed) {
|
||||
System.out.println(bases);
|
||||
displayed = true;
|
||||
}
|
||||
|
||||
System.out.print(base);
|
||||
}
|
||||
else {
|
||||
if (displayed) {
|
||||
System.out.print(" ");
|
||||
}
|
||||
}
|
||||
}
|
||||
if (displayed) {
|
||||
System.out.print("\n");
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
public Integer reduceInit() {
|
||||
return 0;
|
||||
}
|
||||
|
||||
public Integer reduce(Integer value, Integer sum) {
|
||||
return 0;
|
||||
}
|
||||
}
|
||||
|
|
@ -1,52 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers;
|
||||
|
||||
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
|
||||
import org.broadinstitute.sting.gatk.walkers.TreeReducible;
|
||||
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
|
||||
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
|
||||
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
|
||||
import org.broadinstitute.sting.utils.BaseUtils;
|
||||
import org.broadinstitute.sting.utils.BasicPileup;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: chartl
|
||||
* Date: Oct 28, 2009
|
||||
* Time: 3:19:48 PM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
public class IndelCounterWalker extends LocusWalker<Integer,Integer> implements TreeReducible<Integer> {
|
||||
|
||||
public Integer reduceInit() {
|
||||
return 0;
|
||||
}
|
||||
|
||||
public Integer reduce( Integer prevReduce, Integer map ) {
|
||||
return map + prevReduce;
|
||||
}
|
||||
|
||||
public Integer treeReduce(Integer lhs, Integer rhs) {
|
||||
return reduce(lhs,rhs);
|
||||
}
|
||||
|
||||
public Integer map ( RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context ) {
|
||||
if (BaseUtils.isRegularBase(ref.getBase()) ) {
|
||||
return numIndels(context);
|
||||
} else {
|
||||
return 0;
|
||||
}
|
||||
}
|
||||
|
||||
public Integer numIndels ( AlignmentContext context ) {
|
||||
String[] indelPileup = BasicPileup.indelPileup(context.getReads(),context.getOffsets());
|
||||
// number of indels is the number of non-"null" indeces
|
||||
int nIndel = 0;
|
||||
for( String indel : indelPileup ) {
|
||||
if ( ! indel.equals("null") ) { // currently how non-indel bases are represented in pileup
|
||||
nIndel ++;
|
||||
}
|
||||
}
|
||||
|
||||
return nIndel;
|
||||
}
|
||||
}
|
||||
|
|
@ -1,142 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers.Recalibration;
|
||||
|
||||
import org.broadinstitute.sting.gatk.refdata.rodDbSNP;
|
||||
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
|
||||
import org.broadinstitute.sting.gatk.walkers.RMD;
|
||||
import org.broadinstitute.sting.gatk.walkers.DataSource;
|
||||
import org.broadinstitute.sting.gatk.walkers.Requires;
|
||||
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
|
||||
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
|
||||
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
|
||||
import org.broadinstitute.sting.utils.QualityUtils;
|
||||
import org.broadinstitute.sting.utils.Pair;
|
||||
import org.broadinstitute.sting.utils.BaseUtils;
|
||||
import org.broadinstitute.sting.utils.genotype.Variation;
|
||||
import org.broadinstitute.sting.utils.cmdLine.Argument;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.ArrayList;
|
||||
|
||||
import net.sf.samtools.SAMRecord;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: chartl
|
||||
* Date: Oct 20, 2009
|
||||
* Time: 9:41:57 AM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
@Requires(value={DataSource.REFERENCE}, referenceMetaData = {@RMD(name="dbsnp",type= rodDbSNP.class)})
|
||||
public class MinimumNQSWalker extends LocusWalker<Pair<List<Pair<Integer,Integer>>,List<Pair<Integer,Integer>>>, int[][][]> {
|
||||
private static int MM_OFFSET = 1;
|
||||
private static int MATCH_OFFSET = 0;
|
||||
private static int QSCORE_MAX = 1 + QualityUtils.MAX_REASONABLE_Q_SCORE;
|
||||
@Argument(fullName="windowSize", shortName="ws", doc="Size of the window (in one direction)", required=true)
|
||||
private int winSide = 4;
|
||||
|
||||
public void initialize() {
|
||||
out.printf("%s%n", makeHeader());
|
||||
}
|
||||
|
||||
public int[][][] reduceInit() {
|
||||
int[][][] counts = new int[QSCORE_MAX][QSCORE_MAX][2];
|
||||
for ( int i = 0; i < QSCORE_MAX; i ++ ) {
|
||||
for ( int j = 0; j < QSCORE_MAX; j ++ ) {
|
||||
counts[i][j][1]=0;
|
||||
counts[i][j][0]=0;
|
||||
}
|
||||
}
|
||||
return counts;
|
||||
}
|
||||
|
||||
public int[][][] reduce(Pair<List<Pair<Integer,Integer>>,List<Pair<Integer,Integer>>> map, int[][][] prevReduce) {
|
||||
if ( map != null ) {
|
||||
List<Pair<Integer,Integer>> matchingQualityNQSPairs = map.getFirst();
|
||||
List<Pair<Integer,Integer>> mismatchingQualityNQSPairs = map.getSecond();
|
||||
for ( Pair<Integer,Integer> p : matchingQualityNQSPairs ) {
|
||||
prevReduce[p.getFirst()][p.getSecond()][MATCH_OFFSET] ++;
|
||||
}
|
||||
|
||||
for ( Pair<Integer,Integer> p : mismatchingQualityNQSPairs ) {
|
||||
prevReduce[p.getFirst()][p.getSecond()][MM_OFFSET] ++;
|
||||
}
|
||||
}
|
||||
|
||||
return prevReduce;
|
||||
}
|
||||
|
||||
public Pair<List<Pair<Integer,Integer>>,List<Pair<Integer,Integer>>> map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
|
||||
ArrayList<Pair<Integer,Integer>> matchingQualityNQSPairs = new ArrayList<Pair<Integer,Integer>>();
|
||||
ArrayList<Pair<Integer,Integer>> mismatchingQualityNQSPairs = new ArrayList<Pair<Integer,Integer>>();
|
||||
if ( (Variation) tracker.lookup("dbsnp",null) == null ) {
|
||||
for ( int r = 0; r < context.size(); r ++ ) {
|
||||
SAMRecord read = context.getReads().get(r);
|
||||
int offset = context.getOffsets().get(r);
|
||||
int quality = read.getBaseQualities()[offset];
|
||||
int NQS = getNghdQ(offset, read);
|
||||
Pair<Integer,Integer> qualityNQSPair = new Pair<Integer,Integer> (quality,NQS);
|
||||
if ( BaseUtils.basesAreEqual(read.getReadBases()[offset], (byte) ref.getBase()) ) {
|
||||
matchingQualityNQSPairs.add(qualityNQSPair);
|
||||
} else {
|
||||
mismatchingQualityNQSPairs.add(qualityNQSPair);
|
||||
}
|
||||
}
|
||||
|
||||
return new Pair<List<Pair<Integer,Integer>>,List<Pair<Integer,Integer>>>(matchingQualityNQSPairs,mismatchingQualityNQSPairs);
|
||||
} else {
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
public void onTraversalDone( int[][][] reduce ) {
|
||||
for ( int qc = 0; qc < QSCORE_MAX; qc ++ ) {
|
||||
for ( int qn = 0; qn < QSCORE_MAX; qn ++ ) {
|
||||
out.printf("%s%n", formatData(reduce[qc][qn],qc,qn));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
public int getNghdQ(int off, SAMRecord read) {
|
||||
// System.out.println("getNghdQ");
|
||||
byte minQ = Byte.MAX_VALUE;
|
||||
byte[] quals = read.getBaseQualities();
|
||||
int rdlnth = read.getReadLength();
|
||||
int start;
|
||||
int end;
|
||||
if ( off - winSide < 0 ) {
|
||||
start = 0;
|
||||
} else {
|
||||
start = off - winSide;
|
||||
}
|
||||
|
||||
if ( off + winSide > rdlnth ) {
|
||||
end = rdlnth;
|
||||
} else {
|
||||
end = off + winSide;
|
||||
}
|
||||
|
||||
for ( int i = start; i < end; i ++ ) {
|
||||
if ( i != off ) {
|
||||
byte q = quals[i];
|
||||
if ( q < minQ ) {
|
||||
minQ = q;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return minQ;
|
||||
}
|
||||
|
||||
private String makeHeader() {
|
||||
return String.format("%s\t%s\t%s\t%s\t%s\t%s","Reported_Q","Min_Nghd_Q","N_observations","Mm_rate","Empirical_Q","Q_diff");
|
||||
}
|
||||
|
||||
private String formatData( int[] mmArray, int qCenter, int qNghd ) {
|
||||
int counts = mmArray[MM_OFFSET]+mmArray[MATCH_OFFSET];
|
||||
double mismatch = ((double)mmArray[MM_OFFSET]/counts);
|
||||
byte qEmp = QualityUtils.probToQual(1-mismatch);
|
||||
return String.format("%d\t%d\t%d\t%f\t%d\t%d", qCenter, qNghd, counts, mismatch, qEmp,qEmp-qCenter);
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -1,154 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.gatk.walkers.poolseq;
|
||||
|
||||
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
|
||||
import org.broadinstitute.sting.gatk.walkers.genotyper.UnifiedGenotyper;
|
||||
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
|
||||
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
|
||||
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
|
||||
import net.sf.samtools.SAMFileWriter;
|
||||
import net.sf.samtools.SAMRecord;
|
||||
import org.broadinstitute.sting.playground.utils.ArtificialPoolContext;
|
||||
import org.broadinstitute.sting.utils.cmdLine.Argument;
|
||||
import org.broadinstitute.sting.utils.StingException;
|
||||
|
||||
import java.io.FileOutputStream;
|
||||
import java.io.PrintWriter;
|
||||
import java.io.FileNotFoundException;
|
||||
import java.io.IOException;
|
||||
import java.util.Set;
|
||||
import java.util.List;
|
||||
import java.util.ListIterator;
|
||||
import java.util.LinkedList;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: chartl
|
||||
* Date: Aug 26, 2009
|
||||
* Time: 11:28:26 AM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
public class ArtificialPoolWalker extends LocusWalker<ArtificialPoolContext, ArtificialPoolContext> {
|
||||
|
||||
@Argument(fullName = "AuxOutputFile", shortName = "af", doc = "Auxiliary file for genotyp & coverage output", required = true)
|
||||
String auxFilePath = null;
|
||||
@Argument(fullName = "OutputBamFile", shortName = "of", doc = "Output to this file rather than standard output", required = false)
|
||||
SAMFileWriter outputBamFile = null;
|
||||
|
||||
public void initialize() {
|
||||
}
|
||||
|
||||
public ArtificialPoolContext reduceInit() { // try to initialize the file writer
|
||||
ArtificialPoolContext apContext = new ArtificialPoolContext();
|
||||
apContext.setSingleSampleGenotyper(new UnifiedGenotyper());
|
||||
apContext.setReadGroupSets(getToolkit().getMergedReadGroupsByReaders());
|
||||
apContext.setAuxWriter(initializeAuxFileWriter(apContext.getTotalNumberOfPeople()));
|
||||
apContext.setSAMFileWriter(outputBamFile);
|
||||
apContext.initializeUG();
|
||||
|
||||
return apContext;
|
||||
}
|
||||
|
||||
public ArtificialPoolContext map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
|
||||
return new ArtificialPoolContext(tracker,ref,context);
|
||||
}
|
||||
|
||||
public ArtificialPoolContext reduce(ArtificialPoolContext mapCon, ArtificialPoolContext redCon){
|
||||
/* ArtificialPoolContext updatedContext = ArtificialPoolContext.mapReduceMerge(mapCon,redCon);
|
||||
List<SAMRecord>[] newReads = updatedContext.splitReadsByGroup(updatedContext.getNewReads());
|
||||
long[] newCvg = updateRunningCoverage(updatedContext.getRunningCoverage(), getCoverageByGroup(newReads));
|
||||
updatedContext.setRunningCoverage(newCvg);
|
||||
List<SAMRecord>[] sampledReads = ArtificialPoolContext.sampleReads(newReads,runningCoverageToDouble(newCvg));
|
||||
printToFiles(sampledReads,updatedContext);*/
|
||||
AlignmentContext context = redCon.getAlignmentContext();
|
||||
SAMFileWriter writer = redCon.getSAMFileWriter();
|
||||
for(SAMRecord read : context.getReads()) {
|
||||
writer.addAlignment(read);
|
||||
}
|
||||
PrintWriter auxWrite = redCon.getWriterToAuxiliaryFile();
|
||||
auxWrite.print("This is a test.");
|
||||
ArtificialPoolContext updatedContext = redCon;
|
||||
return updatedContext;
|
||||
}
|
||||
|
||||
// Helper methods follow
|
||||
|
||||
private PrintWriter initializeAuxFileWriter(int nFiles) {
|
||||
PrintWriter auxFileWriter;
|
||||
try {
|
||||
auxFileWriter = new PrintWriter(new FileOutputStream(auxFilePath));
|
||||
auxFileWriter.println(createAuxFileHeader(nFiles));
|
||||
} catch(FileNotFoundException e) {
|
||||
String errmsg = "The filepath you entered "+auxFilePath+" could not be opened. Please double-check that the input is correct.";
|
||||
throw new StingException(errmsg, e);
|
||||
} catch(IOException e) {
|
||||
String errmsg = "The file you entered "+auxFilePath+" could not be written to. Please check your permissions to write to this file.";
|
||||
throw new StingException(errmsg,e);
|
||||
}
|
||||
|
||||
return auxFileWriter;
|
||||
}
|
||||
|
||||
private String createAuxFileHeader(int nFiles) {
|
||||
String sp = " ";
|
||||
String st1 = "Chrom:Pos" + sp;
|
||||
String st2 = "";
|
||||
for(int j = 0; j < nFiles; j++) {
|
||||
st2 = st2 + "Pers " + j + " Gen" + sp; // short for "genotype of person j at this location"
|
||||
st2 = st2 + "Pers " + j + " Conf" + sp; // short for "confidence in genotype call of ..."
|
||||
st2 = st2 + "Pers " + j + " NewCvg" + sp; // short for "coverage of person j at this location"
|
||||
}
|
||||
String st3 = "TotalCvg";
|
||||
|
||||
return st1+st2+st3;
|
||||
}
|
||||
|
||||
private int[] getCoverageByGroup(List<SAMRecord>[] readsByGroup) {
|
||||
int[] coverage = new int[readsByGroup.length];
|
||||
for(int iterator = 0; iterator < readsByGroup.length; iterator ++) {
|
||||
coverage[iterator] = readsByGroup[iterator].size();
|
||||
}
|
||||
|
||||
return coverage;
|
||||
}
|
||||
|
||||
private long[] updateRunningCoverage(long[] cvgUpToNow, int[] newCvgByGroup) {
|
||||
long[] newCvg = new long[cvgUpToNow.length];
|
||||
for(int iter = 0; iter < cvgUpToNow.length; iter++) {
|
||||
newCvg[iter] = cvgUpToNow[iter] + newCvgByGroup[iter];
|
||||
}
|
||||
|
||||
return newCvg;
|
||||
}
|
||||
|
||||
private double[] runningCoverageToDouble(long[] cvg) {
|
||||
double[] avgProp = new double[cvg.length];
|
||||
long sum = 0;
|
||||
for(long elem : cvg) {
|
||||
sum += elem;
|
||||
}
|
||||
for(int iter = 0; iter < cvg.length; iter++) {
|
||||
avgProp[iter] = cvg[iter]/sum;
|
||||
}
|
||||
|
||||
return avgProp;
|
||||
}
|
||||
|
||||
private void printToFiles(List<SAMRecord>[] sampledNewReads, ArtificialPoolContext context) {
|
||||
SAMFileWriter samWrite = context.getSAMFileWriter();
|
||||
String sp = " ";
|
||||
PrintWriter auxWrite = context.getWriterToAuxiliaryFile();
|
||||
int readGroupInt = 0;
|
||||
for(List<SAMRecord> readGroup : sampledNewReads) {
|
||||
for(SAMRecord read : readGroup) {
|
||||
samWrite.addAlignment(read);
|
||||
}
|
||||
auxWrite.print(context.getAlignmentContext().getLocation().toString() + sp);
|
||||
auxWrite.print(context.genotypeAndConfidenceToString(readGroupInt,sp));
|
||||
readGroupInt++;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
||||
|
|
@ -1,285 +0,0 @@
|
|||
package org.broadinstitute.sting.playground.utils;
|
||||
|
||||
import net.sf.samtools.SAMFileWriter;
|
||||
import net.sf.samtools.SAMRecord;
|
||||
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
|
||||
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
|
||||
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
|
||||
import org.broadinstitute.sting.gatk.walkers.genotyper.*;
|
||||
import org.broadinstitute.sting.utils.Pair;
|
||||
import org.broadinstitute.sting.utils.genotype.*;
|
||||
|
||||
import java.io.PrintWriter;
|
||||
import java.util.ArrayList;
|
||||
import java.util.LinkedList;
|
||||
import java.util.List;
|
||||
import java.util.Set;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: chartl
|
||||
* Date: Aug 26, 2009
|
||||
* Time: 11:37:42 AM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
public class ArtificialPoolContext {
|
||||
private PrintWriter writerToAuxiliaryFile;
|
||||
private SAMFileWriter writerToSamFile;
|
||||
private UnifiedGenotyper ug;
|
||||
private List<Set<String>> readGroupSets;
|
||||
private long[] runningCoverage;
|
||||
private RefMetaDataTracker refTracker;
|
||||
private ReferenceContext refContext;
|
||||
private AlignmentContext aliContext;
|
||||
|
||||
public ArtificialPoolContext() {
|
||||
readGroupSets = null;
|
||||
writerToAuxiliaryFile = null;
|
||||
writerToSamFile = null;
|
||||
ug = null;
|
||||
refTracker = null;
|
||||
aliContext = null;
|
||||
refContext = null;
|
||||
runningCoverage = null;
|
||||
}
|
||||
|
||||
public ArtificialPoolContext(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
|
||||
refTracker = tracker;
|
||||
refContext = ref;
|
||||
aliContext = context;
|
||||
readGroupSets = null;
|
||||
writerToAuxiliaryFile = null;
|
||||
writerToSamFile=null;
|
||||
ug = null;
|
||||
runningCoverage = null;
|
||||
}
|
||||
|
||||
public ArtificialPoolContext(PrintWriter pw, SAMFileWriter sw, UnifiedGenotyper g, List<Set<String>> rgs, long [] runcvg, RefMetaDataTracker rt, ReferenceContext rc, AlignmentContext ac) {
|
||||
writerToAuxiliaryFile = pw;
|
||||
writerToSamFile = sw;
|
||||
ug = g;
|
||||
readGroupSets = rgs;
|
||||
runningCoverage = runcvg;
|
||||
refTracker = rt;
|
||||
refContext = rc;
|
||||
aliContext = ac;
|
||||
}
|
||||
|
||||
public void setAuxWriter(PrintWriter writer) {
|
||||
writerToAuxiliaryFile = writer;
|
||||
}
|
||||
|
||||
public void setSingleSampleGenotyper(UnifiedGenotyper typer) {
|
||||
ug = typer;
|
||||
}
|
||||
|
||||
public void initializeUG() {
|
||||
ug.initialize();
|
||||
}
|
||||
|
||||
public void setReadGroupSets(List<Set<String>> rgSets) {
|
||||
readGroupSets = rgSets;
|
||||
}
|
||||
|
||||
public void setRefMetaDataTracker(RefMetaDataTracker tracker) {
|
||||
refTracker = tracker;
|
||||
}
|
||||
|
||||
public void setReferenceContext(ReferenceContext ref) {
|
||||
refContext = ref;
|
||||
}
|
||||
|
||||
public void setAlignmentContext(AlignmentContext context) {
|
||||
aliContext = context;
|
||||
}
|
||||
|
||||
public void setRunningCoverage(long[] estimate) {
|
||||
runningCoverage = estimate;
|
||||
}
|
||||
|
||||
public void setSAMFileWriter(SAMFileWriter writer) {
|
||||
writerToSamFile = writer;
|
||||
}
|
||||
|
||||
public int getTotalNumberOfPeople() {
|
||||
return readGroupSets.size();
|
||||
}
|
||||
|
||||
public RefMetaDataTracker getRefMetaDataTracker() {
|
||||
return refTracker;
|
||||
}
|
||||
|
||||
public ReferenceContext getReferenceContext() {
|
||||
return refContext;
|
||||
}
|
||||
|
||||
public AlignmentContext getAlignmentContext() {
|
||||
return aliContext;
|
||||
}
|
||||
|
||||
public PrintWriter getWriterToAuxiliaryFile() {
|
||||
return writerToAuxiliaryFile;
|
||||
}
|
||||
|
||||
public UnifiedGenotyper getSingleSampleGenotyper() {
|
||||
return ug;
|
||||
}
|
||||
|
||||
public List<Set<String>> getReadGroupSets() {
|
||||
return readGroupSets;
|
||||
}
|
||||
|
||||
public long[] getRunningCoverage() {
|
||||
return runningCoverage;
|
||||
}
|
||||
|
||||
public SAMFileWriter getSAMFileWriter() {
|
||||
return writerToSamFile;
|
||||
}
|
||||
|
||||
public List<SAMRecord> getReads() {
|
||||
List<SAMRecord> reads;
|
||||
if(aliContext == null) {
|
||||
reads=null;
|
||||
} else {
|
||||
reads = aliContext.getReads();
|
||||
}
|
||||
|
||||
return reads;
|
||||
}
|
||||
|
||||
public List<Integer> getOffsets() {
|
||||
List<Integer> offsets;
|
||||
if(aliContext == null) {
|
||||
offsets = null;
|
||||
} else {
|
||||
offsets = aliContext.getOffsets();
|
||||
}
|
||||
|
||||
return offsets;
|
||||
}
|
||||
|
||||
public List<SAMRecord> getNewReads() {
|
||||
List<SAMRecord> newReads;
|
||||
|
||||
if(aliContext == null) {
|
||||
newReads = null;
|
||||
} else {
|
||||
newReads = new LinkedList<SAMRecord>();
|
||||
List<SAMRecord> allReads = aliContext.getReads();
|
||||
List<Integer> allOffsets = aliContext.getOffsets();
|
||||
|
||||
for(int iter = 0; iter < allReads.size(); iter++) {
|
||||
if(allOffsets.get(iter) == 0) {
|
||||
newReads.add(allReads.get(iter));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return newReads;
|
||||
}
|
||||
|
||||
public Pair<List<SAMRecord>[],List<Integer>[]> splitByGroup(List<SAMRecord> unsplitReads, List<Integer> unsplitOffsets) {
|
||||
List<SAMRecord>[] readsSplitByGroup;
|
||||
List<Integer> [] offsetsSplitByGroup;
|
||||
|
||||
if(unsplitReads != null && readGroupSets != null) {
|
||||
readsSplitByGroup = new ArrayList[this.getTotalNumberOfPeople()];
|
||||
if(unsplitOffsets != null) {
|
||||
offsetsSplitByGroup = new ArrayList[this.getTotalNumberOfPeople()];
|
||||
}
|
||||
else {
|
||||
offsetsSplitByGroup = null;
|
||||
}
|
||||
int listSize = unsplitReads.size();
|
||||
for(int element = 0; element < listSize; element++) {
|
||||
SAMRecord read = unsplitReads.get(element);
|
||||
for(int groupNumber = 0; groupNumber < this.getTotalNumberOfPeople(); groupNumber++) {
|
||||
if(readGroupSets.get(groupNumber).contains((String) read.getAttribute("RG"))) {
|
||||
readsSplitByGroup[groupNumber].add(read);
|
||||
if(offsetsSplitByGroup != null) {
|
||||
offsetsSplitByGroup[groupNumber].add(unsplitOffsets.get(element));
|
||||
}
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
} else {
|
||||
readsSplitByGroup = null;
|
||||
offsetsSplitByGroup = null; // compiler complains without these lines
|
||||
}
|
||||
|
||||
return new Pair(readsSplitByGroup,offsetsSplitByGroup);
|
||||
}
|
||||
|
||||
public List<SAMRecord>[] splitReadsByGroup(List<SAMRecord> unsplitReads) {
|
||||
return (this.splitByGroup(unsplitReads,null)).first;
|
||||
}
|
||||
|
||||
|
||||
// Static methods follow
|
||||
|
||||
public static ArtificialPoolContext mapReduceMerge(ArtificialPoolContext mapContext, ArtificialPoolContext reduceContext) {
|
||||
return new ArtificialPoolContext(reduceContext.getWriterToAuxiliaryFile(),reduceContext.getSAMFileWriter(),
|
||||
reduceContext.getSingleSampleGenotyper(), reduceContext.getReadGroupSets(), reduceContext.getRunningCoverage(),
|
||||
mapContext.getRefMetaDataTracker(),mapContext.getReferenceContext(),mapContext.getAlignmentContext());
|
||||
}
|
||||
|
||||
public static Pair<List<SAMRecord>[],List<Integer>> sampleReadsAndOffsets(List<SAMRecord>[] reads, List<Integer>[] offsets, double[] propEstGlobal) {
|
||||
double[] samplingRate = calculateSamplingRateFromGlobalEstimate(propEstGlobal);
|
||||
List<SAMRecord>[] sampledReads = new ArrayList[reads.length];
|
||||
List<Integer>[] sampledOffsets;
|
||||
if(offsets != null){
|
||||
sampledOffsets = new ArrayList[offsets.length];
|
||||
} else {
|
||||
sampledOffsets = null;
|
||||
}
|
||||
|
||||
for(int group = 0; group < reads.length; group++) {
|
||||
for(int readNumber = 0; readNumber < reads[group].size(); readNumber++) {
|
||||
if(Math.random() < samplingRate[group]) {
|
||||
sampledReads[group].add(reads[group].get(readNumber));
|
||||
if(sampledOffsets != null) {
|
||||
sampledOffsets[group].add(offsets[group].get(readNumber));
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return new Pair(sampledReads,sampledOffsets);
|
||||
}
|
||||
|
||||
public String genotypeAndConfidenceToString(int group, String spacer) {
|
||||
Genotype call = this.getGenotype(group);
|
||||
return (call.getBases() + spacer + call.getNegLog10PError()); // TODO: fix me
|
||||
}
|
||||
|
||||
public Genotype getGenotype(int group) {
|
||||
AlignmentContext alicon = this.getAlignmentContext();
|
||||
Pair<List<SAMRecord>[],List<Integer>[]> byGroupSplitPair = this.splitByGroup(alicon.getReads(),alicon.getOffsets());
|
||||
Pair<List<Genotype>, GenotypeLocusData> result = ug.map(this.getRefMetaDataTracker(),this.getReferenceContext(),
|
||||
new AlignmentContext(this.getAlignmentContext().getLocation(), byGroupSplitPair.first[group],byGroupSplitPair.second[group]));
|
||||
return (result.first == null ? null : result.first.get(0));
|
||||
}
|
||||
|
||||
public static List<SAMRecord>[] sampleReads(List<SAMRecord>[] reads, double[] propEstGlobal) {
|
||||
return (sampleReadsAndOffsets(reads, null, propEstGlobal)).first;
|
||||
}
|
||||
|
||||
public static double[] calculateSamplingRateFromGlobalEstimate(double[] ratios) {
|
||||
double min = ratios[0];
|
||||
for(double ratio : ratios) {
|
||||
if(ratio < min) {
|
||||
min = ratio;
|
||||
}
|
||||
}
|
||||
double[] samplingRate = new double[ratios.length];
|
||||
// now divide by minimum
|
||||
for(int j = 0; j < ratios.length; j++) {
|
||||
samplingRate[j] = ratios[j]/min;
|
||||
}
|
||||
|
||||
return samplingRate;
|
||||
}
|
||||
|
||||
}
|
||||
|
|
@ -1,43 +0,0 @@
|
|||
/*
|
||||
* The Broad Institute
|
||||
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
|
||||
* This software and its documentation are copyright 2009 by the
|
||||
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
|
||||
*
|
||||
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
|
||||
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
|
||||
*/
|
||||
package org.broadinstitute.sting.playground.utils;
|
||||
|
||||
import edu.mit.broad.picard.util.BasicTextFileParser;
|
||||
|
||||
import java.io.File;
|
||||
import java.util.List;
|
||||
import java.util.Arrays;
|
||||
|
||||
/**
|
||||
* Parser for tab-delimited files
|
||||
*
|
||||
* @author Kathleen Tibbetts
|
||||
*/
|
||||
public class WhitespaceTextFileParser extends BasicTextFileParser {
|
||||
|
||||
/**
|
||||
* Constructor
|
||||
*
|
||||
* @param file The file to parse
|
||||
*/
|
||||
public WhitespaceTextFileParser(boolean treatGroupedDelimitersAsOne, File... file) {
|
||||
super(treatGroupedDelimitersAsOne, file);
|
||||
}
|
||||
|
||||
/**
|
||||
* Determines whether a given character is a delimiter
|
||||
*
|
||||
* @param b the character to evaluate
|
||||
* @return true if <code>b</code> is a delimiter; otherwise false
|
||||
*/
|
||||
protected boolean isDelimiter(byte b) {
|
||||
return Character.isWhitespace((char)b);
|
||||
}
|
||||
}
|
||||
|
|
@ -1,13 +0,0 @@
|
|||
package org.broadinstitute.sting.utils;
|
||||
|
||||
/**
|
||||
* Created by IntelliJ IDEA.
|
||||
* User: depristo
|
||||
* Date: Feb 24, 2009
|
||||
* Time: 10:15:19 AM
|
||||
* To change this template use File | Settings | File Templates.
|
||||
*/
|
||||
public interface Predicate<T> {
|
||||
public boolean apply(T arg);
|
||||
}
|
||||
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
package org.broadinstitute.sting.utils.containers;
|
||||
|
||||
import java.util.PriorityQueue;
|
||||
|
||||
@Deprecated
|
||||
public class BoundedScoringSet<E extends Comparable<E> > {
|
||||
private PriorityQueue<E> pq;
|
||||
private int maximumSize;
|
||||
|
||||
public BoundedScoringSet(int maximumSize) {
|
||||
pq = new PriorityQueue<E>(maximumSize);
|
||||
this.maximumSize = maximumSize;
|
||||
}
|
||||
|
||||
public boolean add(E o) {
|
||||
if (canAdd(o)) {
|
||||
pq.add(o);
|
||||
|
||||
while (pq.size() > maximumSize) {
|
||||
pq.poll();
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
private boolean canAdd(E o) { return pq.size() < maximumSize || o.compareTo(pq.peek()) == 1; }
|
||||
|
||||
public void clear() { pq.clear(); }
|
||||
|
||||
public boolean contains(E o) { return pq.contains(o); }
|
||||
|
||||
public boolean offer(E o) { return pq.offer(o); }
|
||||
|
||||
public E peek() { return pq.peek(); }
|
||||
|
||||
public E poll() { return pq.poll(); }
|
||||
|
||||
public boolean remove(E o) { return pq.remove(o); }
|
||||
|
||||
public int size() { return pq.size(); }
|
||||
|
||||
public E[] toArray(E[] os) { return pq.toArray(os); }
|
||||
}
|
||||
|
|
@ -1,116 +0,0 @@
|
|||
package org.broadinstitute.sting.utils.io;
|
||||
|
||||
import org.broadinstitute.sting.utils.StingException;
|
||||
|
||||
import java.io.OutputStream;
|
||||
import java.io.IOException;
|
||||
import java.io.File;
|
||||
import java.io.FileOutputStream;
|
||||
import java.io.FileNotFoundException;
|
||||
/**
|
||||
* User: hanna
|
||||
* Date: May 26, 2009
|
||||
* Time: 3:51:49 PM
|
||||
* BROAD INSTITUTE SOFTWARE COPYRIGHT NOTICE AND AGREEMENT
|
||||
* Software and documentation are copyright 2005 by the Broad Institute.
|
||||
* All rights are reserved.
|
||||
*
|
||||
* Users acknowledge that this software is supplied without any warranty or support.
|
||||
* The Broad Institute is not responsible for its use, misuse, or
|
||||
* functionality.
|
||||
*/
|
||||
|
||||
/**
|
||||
* An output stream that only initializes itself the first time its used.
|
||||
* Needs a callback that can create an output stream.
|
||||
*/
|
||||
@Deprecated
|
||||
public class LazyFileOutputStream extends OutputStream {
|
||||
/**
|
||||
* Generates output files on demand.
|
||||
*/
|
||||
private final FileFactory factory;
|
||||
|
||||
private File targetFile = null;
|
||||
|
||||
/**
|
||||
* The target for any writes performed by the output stream.
|
||||
*/
|
||||
private FileOutputStream targetOutputStream = null;
|
||||
|
||||
/**
|
||||
* Create a new LazyOutputStream, indicating how to create a new stream.
|
||||
* @param factory Creator of the output stream, when necessary.
|
||||
*/
|
||||
public LazyFileOutputStream( FileFactory factory ) {
|
||||
this.factory = factory;
|
||||
}
|
||||
|
||||
/**
|
||||
* Indicates whether the LazyOutputStream had to get off its butt and create
|
||||
* a new output stream.
|
||||
* @return
|
||||
*/
|
||||
public boolean isCreated() {
|
||||
return targetOutputStream != null;
|
||||
}
|
||||
|
||||
/**
|
||||
* Public method to return the lazily created file.
|
||||
* @return Stream created by the lazy loader.
|
||||
* @throw StingException if no stream was created.
|
||||
*/
|
||||
public File getBackingFile() {
|
||||
if( targetFile == null )
|
||||
throw new StingException("No lazy-loaded stream was created.");
|
||||
return targetFile;
|
||||
}
|
||||
|
||||
@Override
|
||||
public void close() throws IOException {
|
||||
getBackingOutputStream().close();
|
||||
}
|
||||
|
||||
@Override
|
||||
public void flush() throws IOException {
|
||||
getBackingOutputStream().flush();
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(byte[] b) throws IOException {
|
||||
getBackingOutputStream().write(b);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(byte[] b, int off, int len) throws IOException {
|
||||
getBackingOutputStream().write(b,off,len);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(int b) throws IOException {
|
||||
getBackingOutputStream().write(b);
|
||||
}
|
||||
|
||||
/**
|
||||
* Lazy loader for the output stream.
|
||||
*/
|
||||
protected OutputStream getBackingOutputStream() {
|
||||
if( targetOutputStream == null ) {
|
||||
try {
|
||||
targetFile = factory.create();
|
||||
targetOutputStream = new FileOutputStream( targetFile );
|
||||
}
|
||||
catch( IOException ex ) {
|
||||
throw new StingException("Unable to open new temp file", ex );
|
||||
}
|
||||
}
|
||||
return targetOutputStream;
|
||||
}
|
||||
|
||||
/**
|
||||
* Teaches the LazyOutputStream how to create a new outputstream when necessary.
|
||||
*/
|
||||
public interface FileFactory {
|
||||
public File create() throws IOException;
|
||||
}
|
||||
}
|
||||
|
|
@ -1,79 +0,0 @@
|
|||
package org.broadinstitute.sting.utils.io;
|
||||
|
||||
import java.io.OutputStream;
|
||||
import java.io.IOException;
|
||||
/**
|
||||
* User: hanna
|
||||
* Date: Apr 30, 2009
|
||||
* Time: 5:53:32 PM
|
||||
* BROAD INSTITUTE SOFTWARE COPYRIGHT NOTICE AND AGREEMENT
|
||||
* Software and documentation are copyright 2005 by the Broad Institute.
|
||||
* All rights are reserved.
|
||||
*
|
||||
* Users acknowledge that this software is supplied without any warranty or support.
|
||||
* The Broad Institute is not responsible for its use, misuse, or
|
||||
* functionality.
|
||||
*/
|
||||
|
||||
/**
|
||||
* A stream that allows redirection to a variety of sources transparently to the
|
||||
* user of the class.
|
||||
*/
|
||||
@Deprecated
|
||||
public class RedirectingOutputStream extends OutputStream {
|
||||
|
||||
/**
|
||||
* Informs us which output stream should be used.
|
||||
*/
|
||||
private OutputStreamProvider provider;
|
||||
|
||||
/**
|
||||
* Build a new output stream, given the function telling us where to
|
||||
* send output.
|
||||
* @param provider Function which returns an output stream.
|
||||
*/
|
||||
public RedirectingOutputStream( OutputStreamProvider provider ) {
|
||||
this.provider = provider;
|
||||
}
|
||||
|
||||
/**
|
||||
* Gets the OutputStream backing this redirector now.
|
||||
* Note that the backing output stream could change at any time.
|
||||
* Use sparingly (for testing).
|
||||
*/
|
||||
public OutputStream getBackingOutputStream() {
|
||||
return provider.getOutputStream();
|
||||
}
|
||||
|
||||
@Override
|
||||
public void close() throws IOException {
|
||||
getBackingOutputStream().close();
|
||||
}
|
||||
|
||||
@Override
|
||||
public void flush() throws IOException {
|
||||
getBackingOutputStream().flush();
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(byte[] b) throws IOException {
|
||||
getBackingOutputStream().write(b);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(byte[] b, int off, int len) throws IOException {
|
||||
getBackingOutputStream().write(b,off,len);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void write(int b) throws IOException {
|
||||
getBackingOutputStream().write(b);
|
||||
}
|
||||
|
||||
/**
|
||||
* Provides whatever output stream this data should go to at the moment.
|
||||
*/
|
||||
public interface OutputStreamProvider {
|
||||
public OutputStream getOutputStream();
|
||||
}
|
||||
}
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
|
||||
<html>
|
||||
<head>
|
||||
</head>
|
||||
<body bgcolor="white">
|
||||
|
||||
Provides general-purpose tools for handling freeform access to files
|
||||
on disk.
|
||||
|
||||
</body>
|
||||
</html>
|
||||
Loading…
Reference in New Issue