Finish moving VE over to new rod system and fixing up the type inconsistency between eval and comp rods. Now the novel count is always 0 under the known stratification. :)
This commit is contained in:
Eric Banks
parent
90771b74b4
commit
eba316621d
|
|
@ -47,7 +47,8 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
|
|||
public List<RodBinding<VariantContext>> evals;
|
||||
|
||||
@Input(fullName="comp", shortName = "comp", doc="Input comparison file(s)", required=false)
|
||||
public List<RodBinding<VariantContext>> comps = Collections.emptyList();
|
||||
public List<RodBinding<VariantContext>> compsProvided = Collections.emptyList();
|
||||
private List<RodBinding<VariantContext>> comps = new ArrayList<RodBinding<VariantContext>>();
|
||||
|
||||
@ArgumentCollection
|
||||
protected DbsnpArgumentCollection dbsnp = new DbsnpArgumentCollection();
|
||||
|
|
@ -67,7 +68,7 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
|
|||
protected Set<String> SAMPLE_EXPRESSIONS;
|
||||
|
||||
@Argument(shortName="knownName", doc="Name of ROD bindings containing variant sites that should be treated as known when splitting eval rods into known and novel subsets", required=false)
|
||||
protected String[] KNOWN_NAMES = {dbsnp.dbsnp.getName()};
|
||||
protected String[] KNOWN_NAMES = {};
|
||||
|
||||
// Stratification arguments
|
||||
@Argument(fullName="stratificationModule", shortName="ST", doc="One or more specific stratification modules to apply to the eval track(s) (in addition to the standard stratifications, unless -noS is specified)", required=false)
|
||||
|
|
@ -144,11 +145,17 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
|
|||
// Just list the modules, and exit quickly.
|
||||
if (LIST) { variantEvalUtils.listModulesAndExit(); }
|
||||
|
||||
// Add a dummy comp track if none exists
|
||||
if ( comps.size() == 0 ) {
|
||||
comps.add(new RodBinding<VariantContext>(VariantContext.class, "none", "UNBOUND", "", new Tags()));
|
||||
// maintain the full list of comps
|
||||
comps.addAll(compsProvided);
|
||||
if ( dbsnp.dbsnp.isBound() ) {
|
||||
comps.add(dbsnp.dbsnp);
|
||||
knownNames.add(dbsnp.dbsnp.getName());
|
||||
}
|
||||
|
||||
// Add a dummy comp track if none exists
|
||||
if ( comps.size() == 0 )
|
||||
comps.add(new RodBinding<VariantContext>(VariantContext.class, "none", "UNBOUND", "", new Tags()));
|
||||
|
||||
// Cache the rod names
|
||||
for ( RodBinding<VariantContext> compRod : comps )
|
||||
compNames.add(compRod.getName());
|
||||
|
|
@ -156,10 +163,7 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
|
|||
for ( RodBinding<VariantContext> evalRod : evals )
|
||||
evalNames.add(evalRod.getName());
|
||||
|
||||
if ( dbsnp.dbsnp.isBound() )
|
||||
compNames.add(dbsnp.dbsnp.getName());
|
||||
|
||||
// Set up set of known names
|
||||
// Set up set of additional known names
|
||||
knownNames.addAll(Arrays.asList(KNOWN_NAMES));
|
||||
|
||||
// Now that we have all the rods categorized, determine the sample list from the eval rods.
|
||||
|
|
@ -293,23 +297,6 @@ public class VariantEvalWalker extends RodWalker<Integer, Integer> implements Tr
|
|||
}
|
||||
}
|
||||
}
|
||||
|
||||
// todo: Eric, this is really the problem. We select single eval and comp VCs independently
|
||||
// todo: discarding multiple eval tracks at the sites and not providing matched comps
|
||||
// todo: where appropriate. Really this loop should look like:
|
||||
// todo: for each eval track:
|
||||
// todo: for each eval in track:
|
||||
// todo: for each compTrack:
|
||||
// todo: comp = findMatchingComp(eval, compTrack) // find the matching comp in compTrack
|
||||
// todo: call evalModule(eval, comp)
|
||||
// todo: // may return null if no such comp exists, but proceed as eval modules may need to see eval / null pair
|
||||
// todo: for each comp not matched by an eval in compTrack:
|
||||
// todo: call evalModule(null, comp)
|
||||
// todo: // need to call with null comp, as module
|
||||
// todo: note that the reason Kiran pre-computed the possible VCs is to apply the modifiers
|
||||
// todo: like subset to sample, etc. So you probably will want a master map that maps
|
||||
// todo: from special eval bindings to the digested VC for efficiency.
|
||||
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -20,8 +20,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -40,8 +40,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -61,8 +61,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -83,8 +83,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -104,8 +104,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -125,8 +125,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -146,8 +146,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -167,8 +167,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -190,8 +190,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -215,7 +215,7 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:eval,VCF " + fundamentalTestVCF,
|
||||
"--eval " + fundamentalTestVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
@ -232,9 +232,9 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
public void testSelect1() {
|
||||
String extraArgs = "-L 1:1-10,000,000";
|
||||
String tests = cmdRoot +
|
||||
" -B:dbsnp,VCF " + b36dbSNP129 +
|
||||
" -B:eval,VCF3 " + validationDataLocation + "yri.trio.gatk_glftrio.intersection.annotated.filtered.chr1.vcf" +
|
||||
" -B:comp_genotypes,VCF3 " + validationDataLocation + "yri.trio.gatk.ug.head.vcf";
|
||||
" --dbsnp " + b36dbSNP129 +
|
||||
" --eval " + validationDataLocation + "yri.trio.gatk_glftrio.intersection.annotated.filtered.chr1.vcf" +
|
||||
" --comp:comp_genotypes,VCF3 " + validationDataLocation + "yri.trio.gatk.ug.head.vcf";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(withSelect(tests, "DP < 50", "DP50") + " " + extraArgs + " -ST CpG -o %s",
|
||||
1, Arrays.asList("14054badcd89b24c2375e1d09918f681"));
|
||||
executeTestParallel("testSelect1", spec);
|
||||
|
|
@ -244,7 +244,7 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
public void testVEGenotypeConcordance() {
|
||||
String vcfFile = "GenotypeConcordanceEval.vcf";
|
||||
|
||||
WalkerTestSpec spec = new WalkerTestSpec(cmdRoot + " -ST CpG -B:eval,VCF3 " + validationDataLocation + vcfFile + " -B:comp,VCF3 " + validationDataLocation + "GenotypeConcordanceComp.vcf -noEV -EV GenotypeConcordance -o %s",
|
||||
WalkerTestSpec spec = new WalkerTestSpec(cmdRoot + " -ST CpG --eval:VCF3 " + validationDataLocation + vcfFile + " --comp:VCF3 " + validationDataLocation + "GenotypeConcordanceComp.vcf -noEV -EV GenotypeConcordance -o %s",
|
||||
1,
|
||||
Arrays.asList("96f27163f16bb945f19c6623cd6db34e"));
|
||||
executeTestParallel("testVEGenotypeConcordance" + vcfFile, spec);
|
||||
|
|
@ -252,7 +252,7 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
|
||||
@Test
|
||||
public void testCompVsEvalAC() {
|
||||
String extraArgs = "-T VariantEval -R "+b36KGReference+" -o %s -ST CpG -EV GenotypeConcordance -B:evalYRI,VCF3 " + validationDataLocation + "yri.trio.gatk.ug.very.few.lines.vcf -B:compYRI,VCF3 " + validationDataLocation + "yri.trio.gatk.fake.genotypes.ac.test.vcf";
|
||||
String extraArgs = "-T VariantEval -R "+b36KGReference+" -o %s -ST CpG -EV GenotypeConcordance --eval:evalYRI,VCF3 " + validationDataLocation + "yri.trio.gatk.ug.very.few.lines.vcf --comp:compYRI,VCF3 " + validationDataLocation + "yri.trio.gatk.fake.genotypes.ac.test.vcf";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("d1932be3748fcf6da77dc51aec323710"));
|
||||
executeTestParallel("testCompVsEvalAC",spec);
|
||||
}
|
||||
|
|
@ -263,14 +263,14 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
|
||||
@Test
|
||||
public void testTranches() {
|
||||
String extraArgs = "-T VariantEval -R "+ hg18Reference +" -B:eval,vcf " + validationDataLocation + "GA2.WEx.cleaned.ug.snpfiltered.indelfiltered.optimized.vcf -o %s -EV TiTvVariantEvaluator -L chr1 -noEV -ST CpG -tf " + testDir + "tranches.6.txt";
|
||||
String extraArgs = "-T VariantEval -R "+ hg18Reference +" --eval " + validationDataLocation + "GA2.WEx.cleaned.ug.snpfiltered.indelfiltered.optimized.vcf -o %s -EV TiTvVariantEvaluator -L chr1 -noEV -ST CpG -tf " + testDir + "tranches.6.txt";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("984df6e94a546294fc7e0846cbac2dfe"));
|
||||
executeTestParallel("testTranches",spec);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void testCompOverlap() {
|
||||
String extraArgs = "-T VariantEval -R " + b37KGReference + " -L " + validationDataLocation + "VariantEval/pacbio.hg19.intervals -B:comphapmap,vcf " + comparisonDataLocation + "Validated/HapMap/3.3/genotypes_r27_nr.b37_fwd.vcf -B:eval,vcf " + validationDataLocation + "VariantEval/pacbio.ts.recalibrated.vcf -noEV -EV CompOverlap -sn NA12878 -noST -ST Novelty -o %s";
|
||||
String extraArgs = "-T VariantEval -R " + b37KGReference + " -L " + validationDataLocation + "VariantEval/pacbio.hg19.intervals --comp:comphapmap " + comparisonDataLocation + "Validated/HapMap/3.3/genotypes_r27_nr.b37_fwd.vcf --eval " + validationDataLocation + "VariantEval/pacbio.ts.recalibrated.vcf -noEV -EV CompOverlap -sn NA12878 -noST -ST Novelty -o %s";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("462d4784dd55294ef9d5118217b157a5"));
|
||||
executeTestParallel("testCompOverlap",spec);
|
||||
}
|
||||
|
|
@ -280,10 +280,10 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
String extraArgs = "-T VariantEval -R " +
|
||||
b37KGReference +
|
||||
" -L 20" +
|
||||
" -B:dbsnp,VCF " + b37dbSNP132 +
|
||||
" -B:evalBI,VCF " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bi.sites.vcf" +
|
||||
" --dbsnp " + b37dbSNP132 +
|
||||
" --eval:evalBI " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bi.sites.vcf" +
|
||||
" -noST -ST Novelty -o %s";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("0897dfba2f4a245faddce38000555cce"));
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("cbea5f9f8c046d4c014d261db352c43b"));
|
||||
executeTestParallel("testEvalTrackWithoutGenotypes",spec);
|
||||
}
|
||||
|
||||
|
|
@ -291,11 +291,11 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
public void testMultipleEvalTracksWithoutGenotypes() {
|
||||
String extraArgs = "-T VariantEval -R " + b37KGReference +
|
||||
" -L 20" +
|
||||
" -B:dbsnp,VCF " + b37dbSNP132 +
|
||||
" -B:evalBI,VCF " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bi.sites.vcf" +
|
||||
" -B:evalBC,VCF " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bc.sites.vcf" +
|
||||
" --dbsnp " + b37dbSNP132 +
|
||||
" --eval:evalBI " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bi.sites.vcf" +
|
||||
" --eval:evalBC " + validationDataLocation + "VariantEval/ALL.20100201.chr20.bc.sites.vcf" +
|
||||
" -noST -ST Novelty -o %s";
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("ead3602e14ec2944b5d9e4dacc08c819"));
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("d07a246963ae609643620c839b20cd1e"));
|
||||
executeTestParallel("testMultipleEvalTracksWithoutGenotypes",spec);
|
||||
}
|
||||
|
||||
|
|
@ -305,27 +305,27 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
|
||||
String extraArgs = "-T VariantEval" +
|
||||
" -R " + b37KGReference +
|
||||
" -B:comp,VCF " + validationDataLocation + "/VariantEval/ALL.phase1.chr20.broad.snps.genotypes.subset.vcf" +
|
||||
" -B:eval,VCF " + validationDataLocation + "/VariantEval/NA12878.hg19.HiSeq.WGS.cleaned.ug.snpfiltered.indelfiltered.optimized.cut.subset.vcf" +
|
||||
" -B:dbsnp,VCF " + dbsnp +
|
||||
" --comp " + validationDataLocation + "/VariantEval/ALL.phase1.chr20.broad.snps.genotypes.subset.vcf" +
|
||||
" --eval " + validationDataLocation + "/VariantEval/NA12878.hg19.HiSeq.WGS.cleaned.ug.snpfiltered.indelfiltered.optimized.cut.subset.vcf" +
|
||||
" --dbsnp " + dbsnp +
|
||||
" -L 20:10000000-10100000" +
|
||||
" -noST -noEV -ST Novelty -EV CompOverlap" +
|
||||
" -o %s";
|
||||
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("167a347ce0729d1bc3d4fd5069ebd674"));
|
||||
WalkerTestSpec spec = new WalkerTestSpec(extraArgs,1,Arrays.asList("a3c2177849cb00fdff99574cff7f0e4f"));
|
||||
executeTestParallel("testMultipleCompTracks",spec);
|
||||
}
|
||||
|
||||
@Test
|
||||
public void testPerSampleAndSubsettedSampleHaveSameResults() {
|
||||
String md5 = "40471a84b501eb440ee2d42e3081f228";
|
||||
String md5 = "dab415cc76846e18fcf8c78f2b2ee033";
|
||||
|
||||
WalkerTestSpec spec = new WalkerTestSpec(
|
||||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestSNPsVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestSNPsVCF,
|
||||
"-noEV",
|
||||
"-EV CompOverlap",
|
||||
"-sn HG00625",
|
||||
|
|
@ -342,8 +342,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestSNPsOneSampleVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestSNPsOneSampleVCF,
|
||||
"-noEV",
|
||||
"-EV CompOverlap",
|
||||
"-noST",
|
||||
|
|
@ -363,8 +363,8 @@ public class VariantEvalIntegrationTest extends WalkerTest {
|
|||
buildCommandLine(
|
||||
"-T VariantEval",
|
||||
"-R " + b37KGReference,
|
||||
"-B:dbsnp,VCF " + b37dbSNP132,
|
||||
"-B:eval,VCF " + fundamentalTestSNPsVCF,
|
||||
"--dbsnp " + b37dbSNP132,
|
||||
"--eval " + fundamentalTestSNPsVCF,
|
||||
"-noEV",
|
||||
"-EV CountVariants",
|
||||
"-noST",
|
||||
|
|
|
|||
Loading…
Reference in New Issue