Moving GAV walker to public

Walker is updated to the new RodBinding system and has the new GATKDocs layout.
This commit is contained in:
Mauricio Carneiro 2011-08-17 21:54:40 -04:00
parent a9df365364
commit cc3df8f11a
1 changed files with 438 additions and 0 deletions

View File

@ -0,0 +1,438 @@
/*
* Copyright (c) 2010 The Broad Institute
*
* Permission is hereby granted, free of charge, to any person
* obtaining a copy of this software and associated documentation
* files (the "Software"), to deal in the Software without
* restriction, including without limitation the rights to use,
* copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following
* conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
* OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
* HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
* WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
* THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
package org.broadinstitute.sting.gatk.walkers.validation;
import org.broadinstitute.sting.commandline.Argument;
import org.broadinstitute.sting.commandline.Input;
import org.broadinstitute.sting.commandline.Output;
import org.broadinstitute.sting.commandline.RodBinding;
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.*;
import org.broadinstitute.sting.gatk.walkers.genotyper.GenotypeLikelihoodsCalculationModel;
import org.broadinstitute.sting.gatk.walkers.genotyper.UnifiedArgumentCollection;
import org.broadinstitute.sting.gatk.walkers.genotyper.UnifiedGenotyperEngine;
import org.broadinstitute.sting.gatk.walkers.genotyper.VariantCallContext;
import org.broadinstitute.sting.utils.SampleUtils;
import org.broadinstitute.sting.utils.codecs.vcf.VCFHeader;
import org.broadinstitute.sting.utils.codecs.vcf.VCFHeaderLine;
import org.broadinstitute.sting.utils.codecs.vcf.VCFUtils;
import org.broadinstitute.sting.utils.codecs.vcf.VCFWriter;
import org.broadinstitute.sting.utils.exceptions.UserException;
import org.broadinstitute.sting.utils.variantcontext.MutableVariantContext;
import org.broadinstitute.sting.utils.variantcontext.VariantContext;
import org.broadinstitute.sting.utils.variantcontext.VariantContextUtils;
import java.util.Map;
import java.util.Set;
import static org.broadinstitute.sting.utils.IndelUtils.isInsideExtendedIndel;
/**
* Genotypes a dataset and validates the calls of another dataset using the Unified Genotyper.
*
* <p>
* Genotype and Validate is a tool to evaluate the quality of a dataset for calling SNPs
* and Indels given a secondary (validation) data source. The data sources are BAM or VCF
* files. You can use them interchangeably (i.e. a BAM to validate calls in a VCF or a VCF
* to validate calls on a BAM).
* </p>
*
* <p>
* The simplest scenario is when you have a VCF of hand annotated SNPs and Indels, and you
* want to know how well a particular technology performs calling these snps. With a
* dataset (BAM file) generated by the technology in test, and the hand annotated VCF, you
* can run GenotypeAndValidate to asses the accuracy of the calls with the new technology's
* dataset.
* </p>
*
* <p>
* Another option is to validate the calls on a VCF file, using a deep coverage BAM file
* that you trust the calls on. The GenotypeAndValidate walker will make calls using the
* reads in the BAM file and take them as truth, then compare to the calls in the VCF file
* and produce a truth table.
* </p>
*
*
* <h2>Input</h2>
* <p>
* A BAM file to make calls on and a VCF file to use as truth validation dataset.
*
* You also have the option to invert the roles of the files using the command line options listed below.
* </p>
*
* <h2>Output</h2>
* <p>
* GenotypeAndValidate has two outputs. The truth table and the optional VCF file. The truth table is a
* 2x2 table correlating what was called in the dataset with the truth of the call (whether it's a true
* positive or a false positive). The table should look like this:
* </p>
* <center>
* <table id="description-table">
* <tr>
* <th></th>
* <th>ALT</th>
* <th>REF</th>
* <th>Predictive Value</th>
* </tr>
* <tr>
* <td><b>called alt</b></td>
* <td>True Positive (TP)</td>
* <td>False Positive (FP)</td>
* <td>Positive PV</td>
* </tr>
* <tr>
* <td><b>called ref</b></td>
* <td>False Negative (FN)</td>
* <td>True Negative (TN)</td>
* <td>Negative PV</td>
* </tr>
* </table>
* </center>
*
* <p>
* The <b>positive predictive value (PPV)</b> is the proportion of subjects with positive test results
* who are correctly diagnosed.
* </p>
* <p>
* The <b>negative predictive value (NPV)</b> is the proportion of subjects with a negative test result
* who are correctly diagnosed.
* </p>
* <p>
* The VCF file will contain only the variants that were called or not called, excluding the ones that
* were uncovered or didn't pass the filters. This file is useful if you are trying to compare
* the PPV and NPV of two different technologies on the exact same sites (so you can compare apples to
* apples).
* </p>
*
* <p>
* Here is an example of an annotated VCF file (info field clipped for clarity)
*
* <pre>
* #CHROM POS ID REF ALT QUAL FILTER INFO FORMAT NA12878
* 1 20568807 . C T 0 HapMapHet AC=1;AF=0.50;AN=2;DP=0;GV=T GT 0/1
* 1 22359922 . T C 282 WG-CG-HiSeq AC=2;AF=0.50;GV=T;AN=4;DP=42 GT:AD:DP:GL:GQ 1/0 ./. 0/1:20,22:39:-72.79,-11.75,-67.94:99 ./.
* 13 102391461 . G A 341 Indel;SnpCluster AC=1;GV=F;AF=0.50;AN=2;DP=45 GT:AD:DP:GL:GQ ./. ./. 0/1:32,13:45:-50.99,-13.56,-112.17:99 ./.
* 1 175516757 . C G 655 SnpCluster,WG AC=1;AF=0.50;AN=2;GV=F;DP=74 GT:AD:DP:GL:GQ ./. ./. 0/1:52,22:67:-89.02,-20.20,-191.27:99 ./.
* </pre>
*
* </p>
*
* <h3>Additional Details</h3>
* <ul>
* <li>
* You should always use -BTI on your VCF track, so that the GATK only looks at the sites on the VCF file.
* This speeds up the process a lot.
* </li>
* <li>
* The total number of visited bases may be greater than the number of variants in the original
* VCF file because of extended indels, as they trigger one call per new insertion or deletion.
* (i.e. ACTG/- will count as 4 genotyper calls, but it's only one line in the VCF).
* </li>
* </ul>
*
* <h2>Examples</h2>
* <ol>
* <li>
* Genotypes BAM file from new technology using the VCF as a truth dataset:
* </li>
*
* <pre>
* java
* -jar /GenomeAnalysisTK.jar
* -T GenotypeAndValidate
* -R human_g1k_v37.fasta
* -I myNewTechReads.bam
* -alleles handAnnotatedVCF.vcf
* -BTI alleles
* </pre>
*
* <li>
* Using a BAM file as the truth dataset:
* </li>
*
* <pre>
* java
* -jar /GenomeAnalysisTK.jar
* -T GenotypeAndValidate
* -R human_g1k_v37.fasta
* -I myTruthDataset.bam
* -alleles callsToValidate.vcf
* -BTI alleles
* -bt
* -o gav.vcf
* </pre>
*
*
* @author Mauricio Carneiro
* @since ${DATE}
*/
@Requires(value={DataSource.READS, DataSource.REFERENCE})
@Allows(value={DataSource.READS, DataSource.REFERENCE})
@By(DataSource.REFERENCE)
@Reference(window=@Window(start=-200,stop=200))
public class GenotypeAndValidateWalker extends RodWalker<GenotypeAndValidateWalker.CountedData, GenotypeAndValidateWalker.CountedData> implements TreeReducible<GenotypeAndValidateWalker.CountedData> {
@Output(doc="Generate a VCF file with the variants considered by the walker, with a new annotation \"callStatus\" which will carry the value called in the validation VCF or BAM file", required=false)
protected VCFWriter vcfWriter = null;
@Input(fullName="alleles", shortName = "alleles", doc="The set of alleles at which to genotype", required=true)
public RodBinding<VariantContext> alleles;
@Argument(fullName ="set_bam_truth", shortName ="bt", doc="Use the calls on the reads (bam file) as the truth dataset and validate the calls on the VCF", required=false)
private boolean bamIsTruth = false;
@Argument(fullName="minimum_base_quality_score", shortName="mbq", doc="Minimum base quality score for calling a genotype", required=false)
private int mbq = -1;
@Argument(fullName="maximum_deletion_fraction", shortName="deletions", doc="Maximum deletion fraction for calling a genotype", required=false)
private double deletions = -1;
@Argument(fullName="standard_min_confidence_threshold_for_calling", shortName="stand_call_conf", doc="the minimum phred-scaled Qscore threshold to separate high confidence from low confidence calls", required=false)
private double callConf = -1;
@Argument(fullName="standard_min_confidence_threshold_for_emitting", shortName="stand_emit_conf", doc="the minimum phred-scaled Qscore threshold to emit low confidence calls", required=false)
private double emitConf = -1;
@Argument(fullName="condition_on_depth", shortName="depth", doc="Condition validation on a minimum depth of coverage by the reads", required=false)
private int minDepth = -1;
@Argument(fullName ="sample", shortName ="sn", doc="Name of the sample to validate (in case your VCF/BAM has more than one sample)", required=false)
private String sample = "";
private UnifiedGenotyperEngine snpEngine;
private UnifiedGenotyperEngine indelEngine;
public static class CountedData {
private long nAltCalledAlt = 0L;
private long nAltCalledRef = 0L;
private long nRefCalledAlt = 0L;
private long nRefCalledRef = 0L;
private long nNotConfidentCalls = 0L;
private long nUncovered = 0L;
/**
* Adds the values of other to this, returning this
* @param other the other object
*/
public void add(CountedData other) {
nAltCalledAlt += other.nAltCalledAlt;
nAltCalledRef += other.nAltCalledRef;
nRefCalledAlt += other.nRefCalledAlt;
nRefCalledRef += other.nRefCalledRef;
nUncovered += other.nUncovered;
nNotConfidentCalls += other.nNotConfidentCalls;
}
}
//---------------------------------------------------------------------------------------------------------------
//
// initialize
//
//---------------------------------------------------------------------------------------------------------------
public void initialize() {
// Initialize VCF header
if (vcfWriter != null) {
Map<String, VCFHeader> header = VCFUtils.getVCFHeadersFromRodPrefix(getToolkit(), alleles.getName());
Set<String> samples = SampleUtils.getSampleList(header, VariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE);
Set<VCFHeaderLine> headerLines = VCFUtils.smartMergeHeaders(header.values(), logger);
headerLines.add(new VCFHeaderLine("source", "GenotypeAndValidate"));
vcfWriter.writeHeader(new VCFHeader(headerLines, samples));
}
// Filling in SNP calling arguments for UG
UnifiedArgumentCollection uac = new UnifiedArgumentCollection();
uac.OutputMode = UnifiedGenotyperEngine.OUTPUT_MODE.EMIT_ALL_SITES;
uac.alleles = alleles;
if (!bamIsTruth) uac.GenotypingMode = GenotypeLikelihoodsCalculationModel.GENOTYPING_MODE.GENOTYPE_GIVEN_ALLELES;
if (mbq >= 0) uac.MIN_BASE_QUALTY_SCORE = mbq;
if (deletions >= 0) uac.MAX_DELETION_FRACTION = deletions;
if (emitConf >= 0) uac.STANDARD_CONFIDENCE_FOR_EMITTING = emitConf;
if (callConf >= 0) uac.STANDARD_CONFIDENCE_FOR_CALLING = callConf;
uac.GLmodel = GenotypeLikelihoodsCalculationModel.Model.SNP;
snpEngine = new UnifiedGenotyperEngine(getToolkit(), uac);
// Adding the INDEL calling arguments for UG
uac.GLmodel = GenotypeLikelihoodsCalculationModel.Model.INDEL;
indelEngine = new UnifiedGenotyperEngine(getToolkit(), uac);
// make sure we have callConf set to the threshold set by the UAC so we can use it later.
callConf = uac.STANDARD_CONFIDENCE_FOR_CALLING;
}
//---------------------------------------------------------------------------------------------------------------
//
// map
//
//---------------------------------------------------------------------------------------------------------------
public CountedData map( RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context ) {
final CountedData counter = new CountedData();
// For some reason RodWalkers get map calls with null trackers
if( tracker == null )
return counter;
VariantContext vcComp = tracker.getFirstValue(alleles);
if( vcComp == null )
return counter;
//todo - not sure I want this, may be misleading to filter extended indel events.
if (isInsideExtendedIndel(vcComp, ref))
return counter;
// Do not operate on variants that are not covered to the optional minimum depth
if (!context.hasReads() || (minDepth > 0 && context.getBasePileup().getBases().length < minDepth)) {
counter.nUncovered = 1L;
return counter;
}
VariantCallContext call;
if ( vcComp.isSNP() )
call = snpEngine.calculateLikelihoodsAndGenotypes(tracker, ref, context);
else if ( vcComp.isIndel() ) {
call = indelEngine.calculateLikelihoodsAndGenotypes(tracker, ref, context);
}
else {
logger.info("Not SNP or INDEL " + vcComp.getChr() + ":" + vcComp.getStart() + " " + vcComp.getAlleles());
return counter;
}
boolean writeVariant = true;
if (bamIsTruth) {
if (call.confidentlyCalled) {
// If truth is a confident REF call
if (call.isVariant()) {
if (vcComp.isVariant())
counter.nAltCalledAlt = 1L; // todo -- may wanna check if the alts called are the same?
else
counter.nAltCalledRef = 1L;
}
// If truth is a confident ALT call
else {
if (vcComp.isVariant())
counter.nRefCalledAlt = 1L;
else
counter.nRefCalledRef = 1L;
}
}
else {
counter.nNotConfidentCalls = 1L;
writeVariant = false;
}
}
else {
if (!vcComp.hasAttribute("GV"))
throw new UserException.BadInput("Variant has no GV annotation in the INFO field. " + vcComp.getChr() + ":" + vcComp.getStart());
if (call.isCalledAlt(callConf)) {
if (vcComp.getAttribute("GV").equals("T"))
counter.nAltCalledAlt = 1L;
else
counter.nRefCalledAlt = 1L;
}
else if (call.isCalledRef(callConf)) {
if (vcComp.getAttribute("GV").equals("T"))
counter.nAltCalledRef = 1L;
else
counter.nRefCalledRef = 1L;
}
else {
counter.nNotConfidentCalls = 1L;
writeVariant = false;
}
}
if (vcfWriter != null && writeVariant) {
if (!vcComp.hasAttribute("callStatus")) {
MutableVariantContext mvc = new MutableVariantContext(vcComp);
mvc.putAttribute("callStatus", call.isCalledAlt(callConf) ? "ALT" : "REF" );
vcfWriter.add(mvc);
}
else
vcfWriter.add(vcComp);
}
return counter;
}
//---------------------------------------------------------------------------------------------------------------
//
// reduce
//
//---------------------------------------------------------------------------------------------------------------
public CountedData reduceInit() {
return new CountedData();
}
public CountedData treeReduce( final CountedData sum1, final CountedData sum2) {
sum2.add(sum1);
return sum2;
}
public CountedData reduce( final CountedData mapValue, final CountedData reduceSum ) {
reduceSum.add(mapValue);
return reduceSum;
}
public void onTraversalDone( CountedData reduceSum ) {
double ppv = 100 * ((double) reduceSum.nAltCalledAlt /( reduceSum.nAltCalledAlt + reduceSum.nRefCalledAlt));
double npv = 100 * ((double) reduceSum.nRefCalledRef /( reduceSum.nRefCalledRef + reduceSum.nAltCalledRef));
double sensitivity = 100 * ((double) reduceSum.nAltCalledAlt /( reduceSum.nAltCalledAlt + reduceSum.nAltCalledRef));
double specificity = (reduceSum.nRefCalledRef + reduceSum.nRefCalledAlt > 0) ? 100 * ((double) reduceSum.nRefCalledRef /( reduceSum.nRefCalledRef + reduceSum.nRefCalledAlt)) : 100;
logger.info(String.format("Resulting Truth Table Output\n\n" +
"---------------------------------------------------\n" +
"\t\t|\tALT\t|\tREF\t\n" +
"---------------------------------------------------\n" +
"called alt\t|\t%d\t|\t%d\n" +
"called ref\t|\t%d\t|\t%d\n" +
"---------------------------------------------------\n" +
"positive predictive value: %f%%\n" +
"negative predictive value: %f%%\n" +
"---------------------------------------------------\n" +
"sensitivity: %f%%\n" +
"specificity: %f%%\n" +
"---------------------------------------------------\n" +
"not confident: %d\n" +
"not covered: %d\n" +
"---------------------------------------------------\n", reduceSum.nAltCalledAlt, reduceSum.nRefCalledAlt, reduceSum.nAltCalledRef, reduceSum.nRefCalledRef, ppv, npv, sensitivity, specificity, reduceSum.nNotConfidentCalls, reduceSum.nUncovered));
}
}