public static void addComplexGenotypesTest() {
final List<Allele> allAlleles =
Arrays.asList(
Allele.create("A", true), Allele.create("C", false), Allele.create("G", false));
for (int nAlleles : Arrays.asList(2, 3)) {
for (int highestPloidy : Arrays.asList(1, 2, 3)) {
// site alleles
final List<Allele> siteAlleles = allAlleles.subList(0, nAlleles);
// possible alleles for genotypes
final List<Allele> possibleGenotypeAlleles = new ArrayList<Allele>(siteAlleles);
possibleGenotypeAlleles.add(Allele.NO_CALL);
// there are n^ploidy possible genotypes
final List<List<Allele>> possibleGenotypes =
makeAllGenotypes(possibleGenotypeAlleles, highestPloidy);
final int nPossibleGenotypes = possibleGenotypes.size();
VariantContextBuilder vb = new VariantContextBuilder("unittest", "1", 1, 1, siteAlleles);
// first test -- create n copies of each genotype
for (int i = 0; i < nPossibleGenotypes; i++) {
final List<Genotype> samples = new ArrayList<Genotype>(3);
samples.add(GenotypeBuilder.create("sample" + i, possibleGenotypes.get(i)));
add(vb.genotypes(samples));
}
// second test -- create one sample with each genotype
{
final List<Genotype> samples = new ArrayList<Genotype>(nPossibleGenotypes);
for (int i = 0; i < nPossibleGenotypes; i++) {
samples.add(GenotypeBuilder.create("sample" + i, possibleGenotypes.get(i)));
}
add(vb.genotypes(samples));
}
// test mixed ploidy
for (int i = 0; i < nPossibleGenotypes; i++) {
for (int ploidy = 1; ploidy < highestPloidy; ploidy++) {
final List<Genotype> samples = new ArrayList<Genotype>(highestPloidy);
final List<Allele> genotype = possibleGenotypes.get(i).subList(0, ploidy);
samples.add(GenotypeBuilder.create("sample" + i, genotype));
add(vb.genotypes(samples));
}
}
}
}
}
private static Genotype attr(
final String name, final Allele ref, final String key, final Object... value) {
if (value.length == 0) return GenotypeBuilder.create(name, Arrays.asList(ref, ref));
else {
final Object toAdd = value.length == 1 ? value[0] : Arrays.asList(value);
return new GenotypeBuilder(name, Arrays.asList(ref, ref)).attribute(key, toAdd).make();
}
}
/**
* Takes the interval, finds it in the stash, prints it to the VCF
*
* @param stats The statistics of the interval
* @param refAllele the reference allele
*/
private void outputStatsToVCF(final IntervalStratification stats, final Allele refAllele) {
GenomeLoc interval = stats.getInterval();
final List<Allele> alleles = new ArrayList<>();
final Map<String, Object> attributes = new HashMap<>();
final ArrayList<Genotype> genotypes = new ArrayList<>();
for (String sample : samples) {
final GenotypeBuilder gb = new GenotypeBuilder(sample);
SampleStratification sampleStat = stats.getSampleStatistics(sample);
gb.attribute(
GATKVCFConstants.AVG_INTERVAL_DP_BY_SAMPLE_KEY,
sampleStat.averageCoverage(interval.size()));
gb.attribute(GATKVCFConstants.LOW_COVERAGE_LOCI, sampleStat.getNLowCoveredLoci());
gb.attribute(GATKVCFConstants.ZERO_COVERAGE_LOCI, sampleStat.getNUncoveredLoci());
gb.filters(statusToStrings(stats.getSampleStatistics(sample).callableStatuses(), false));
genotypes.add(gb.make());
}
alleles.add(refAllele);
alleles.add(SYMBOLIC_ALLELE);
VariantContextBuilder vcb =
new VariantContextBuilder(
"DiagnoseTargets",
interval.getContig(),
interval.getStart(),
interval.getStop(),
alleles);
vcb = vcb.log10PError(VariantContext.NO_LOG10_PERROR);
vcb.filters(new LinkedHashSet<>(statusToStrings(stats.callableStatuses(), true)));
attributes.put(VCFConstants.END_KEY, interval.getStop());
attributes.put(GATKVCFConstants.AVG_INTERVAL_DP_KEY, stats.averageCoverage(interval.size()));
attributes.put(GATKVCFConstants.INTERVAL_GC_CONTENT_KEY, stats.gcContent());
vcb = vcb.attributes(attributes);
vcb = vcb.genotypes(genotypes);
vcfWriter.add(vcb.make());
}
private static void addGenotypesAndGTests() {
// for ( final int ploidy : Arrays.asList(2)) {
for (final int ploidy : Arrays.asList(1, 2, 3, 4, 5)) {
final List<List<String>> alleleCombinations =
Arrays.asList(
Arrays.asList("A"),
Arrays.asList("A", "C"),
Arrays.asList("A", "C", "G"),
Arrays.asList("A", "C", "G", "T"));
for (final List<String> alleles : alleleCombinations) {
final VariantContextBuilder vcb = builder().alleles(alleles);
final VariantContext site = vcb.make();
final int nAlleles = site.getNAlleles();
final Allele ref = site.getReference();
// base genotype is ref/.../ref up to ploidy
final List<Allele> baseGenotype = new ArrayList<Allele>(ploidy);
for (int i = 0; i < ploidy; i++) baseGenotype.add(ref);
final int nPLs = GenotypeLikelihoods.numLikelihoods(nAlleles, ploidy);
// ada is 0, 1, ..., nAlleles - 1
final List<Integer> ada = new ArrayList<Integer>(nAlleles);
for (int i = 0; i < nAlleles - 1; i++) ada.add(i);
// pl is 0, 1, ..., up to nPLs (complex calc of nAlleles and ploidy)
final int[] pl = new int[nPLs];
for (int i = 0; i < pl.length; i++) pl[i] = i;
final GenotypeBuilder gb = new GenotypeBuilder("ADA_PL_SAMPLE");
gb.alleles(baseGenotype);
gb.PL(pl);
gb.attribute("ADA", nAlleles == 2 ? ada.get(0) : ada);
vcb.genotypes(gb.make());
add(vcb);
}
}
}
private static void addGenotypes(final VariantContext site) {
// test ref/ref
final Allele ref = site.getReference();
final Allele alt1 = site.getNAlleles() > 1 ? site.getAlternateAllele(0) : null;
final Genotype homRef = GenotypeBuilder.create("homRef", Arrays.asList(ref, ref));
addGenotypeTests(site, homRef);
if (alt1 != null) {
final Genotype het = GenotypeBuilder.create("het", Arrays.asList(ref, alt1));
final Genotype homVar = GenotypeBuilder.create("homVar", Arrays.asList(alt1, alt1));
addGenotypeTests(site, homRef, het);
addGenotypeTests(site, homRef, het, homVar);
// test no GT at all
addGenotypeTests(
site, new GenotypeBuilder("noGT", new ArrayList<Allele>(0)).attribute("INT1", 10).make());
final List<Allele> noCall = Arrays.asList(Allele.NO_CALL, Allele.NO_CALL);
// ploidy
if (ENABLE_PLOIDY_TESTS) {
addGenotypeTests(
site,
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("hap", Arrays.asList(ref)));
addGenotypeTests(
site,
GenotypeBuilder.create("noCall", noCall),
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("hap", Arrays.asList(ref)));
addGenotypeTests(
site,
GenotypeBuilder.create("noCall", noCall),
GenotypeBuilder.create("noCall2", noCall),
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("hap", Arrays.asList(ref)));
addGenotypeTests(
site,
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("tet", Arrays.asList(ref, alt1, alt1)));
addGenotypeTests(
site,
GenotypeBuilder.create("noCall", noCall),
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("tet", Arrays.asList(ref, alt1, alt1)));
addGenotypeTests(
site,
GenotypeBuilder.create("noCall", noCall),
GenotypeBuilder.create("noCall2", noCall),
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("tet", Arrays.asList(ref, alt1, alt1)));
addGenotypeTests(
site,
GenotypeBuilder.create("nocall", noCall),
GenotypeBuilder.create("dip", Arrays.asList(ref, alt1)),
GenotypeBuilder.create("tet", Arrays.asList(ref, alt1, alt1)));
}
//
//
// TESTING PHASE
//
//
final Genotype gUnphased = new GenotypeBuilder("gUnphased", Arrays.asList(ref, alt1)).make();
final Genotype gPhased =
new GenotypeBuilder("gPhased", Arrays.asList(ref, alt1)).phased(true).make();
final Genotype gPhased2 =
new GenotypeBuilder("gPhased2", Arrays.asList(alt1, alt1)).phased(true).make();
final Genotype gPhased3 =
new GenotypeBuilder("gPhased3", Arrays.asList(ref, ref)).phased(true).make();
final Genotype haploidNoPhase =
new GenotypeBuilder("haploidNoPhase", Arrays.asList(ref)).make();
addGenotypeTests(site, gUnphased, gPhased);
addGenotypeTests(site, gUnphased, gPhased2);
addGenotypeTests(site, gUnphased, gPhased3);
addGenotypeTests(site, gPhased, gPhased2);
addGenotypeTests(site, gPhased, gPhased3);
addGenotypeTests(site, gPhased2, gPhased3);
addGenotypeTests(site, haploidNoPhase, gPhased);
addGenotypeTests(site, haploidNoPhase, gPhased2);
addGenotypeTests(site, haploidNoPhase, gPhased3);
addGenotypeTests(site, haploidNoPhase, gPhased, gPhased2);
addGenotypeTests(site, haploidNoPhase, gPhased, gPhased3);
addGenotypeTests(site, haploidNoPhase, gPhased2, gPhased3);
addGenotypeTests(site, haploidNoPhase, gPhased, gPhased2, gPhased3);
final Genotype gUnphasedTet =
new GenotypeBuilder("gUnphasedTet", Arrays.asList(ref, alt1, ref, alt1)).make();
final Genotype gPhasedTet =
new GenotypeBuilder("gPhasedTet", Arrays.asList(ref, alt1, alt1, alt1))
.phased(true)
.make();
addGenotypeTests(site, gUnphasedTet, gPhasedTet);
}
if (ENABLE_PL_TESTS) {
if (site.getNAlleles() == 2) {
// testing PLs
addGenotypeTests(
site,
GenotypeBuilder.create("g1", Arrays.asList(ref, ref), new double[] {0, -1, -2}),
GenotypeBuilder.create("g2", Arrays.asList(ref, ref), new double[] {0, -2, -3}));
addGenotypeTests(
site,
GenotypeBuilder.create("g1", Arrays.asList(ref, ref), new double[] {-1, 0, -2}),
GenotypeBuilder.create("g2", Arrays.asList(ref, ref), new double[] {0, -2, -3}));
addGenotypeTests(
site,
GenotypeBuilder.create("g1", Arrays.asList(ref, ref), new double[] {-1, 0, -2}),
GenotypeBuilder.create("g2", Arrays.asList(ref, ref), new double[] {0, -2000, -1000}));
addGenotypeTests(
site, // missing PLs
GenotypeBuilder.create("g1", Arrays.asList(ref, ref), new double[] {-1, 0, -2}),
GenotypeBuilder.create("g2", Arrays.asList(ref, ref)));
} else if (site.getNAlleles() == 3) {
// testing PLs
addGenotypeTests(
site,
GenotypeBuilder.create(
"g1", Arrays.asList(ref, ref), new double[] {0, -1, -2, -3, -4, -5}),
GenotypeBuilder.create(
"g2", Arrays.asList(ref, ref), new double[] {0, -2, -3, -4, -5, -6}));
}
}
// test attributes
addGenotypeTests(site, attr("g1", ref, "INT1", 1), attr("g2", ref, "INT1", 2));
addGenotypeTests(site, attr("g1", ref, "INT1", 1), attr("g2", ref, "INT1"));
addGenotypeTests(site, attr("g1", ref, "INT3", 1, 2, 3), attr("g2", ref, "INT3", 4, 5, 6));
addGenotypeTests(site, attr("g1", ref, "INT3", 1, 2, 3), attr("g2", ref, "INT3"));
addGenotypeTests(
site, attr("g1", ref, "INT20", TWENTY_INTS), attr("g2", ref, "INT20", TWENTY_INTS));
if (ENABLE_VARARRAY_TESTS) {
addGenotypeTests(
site,
attr("g1", ref, "INT.VAR", 1, 2, 3),
attr("g2", ref, "INT.VAR", 4, 5),
attr("g3", ref, "INT.VAR", 6));
addGenotypeTests(
site,
attr("g1", ref, "INT.VAR", 1, 2, 3),
attr("g2", ref, "INT.VAR"),
attr("g3", ref, "INT.VAR", 5));
}
addGenotypeTests(site, attr("g1", ref, "FLOAT1", 1.0), attr("g2", ref, "FLOAT1", 2.0));
addGenotypeTests(site, attr("g1", ref, "FLOAT1", 1.0), attr("g2", ref, "FLOAT1"));
addGenotypeTests(
site, attr("g1", ref, "FLOAT3", 1.0, 2.0, 3.0), attr("g2", ref, "FLOAT3", 4.0, 5.0, 6.0));
addGenotypeTests(site, attr("g1", ref, "FLOAT3", 1.0, 2.0, 3.0), attr("g2", ref, "FLOAT3"));
if (ENABLE_VARIABLE_LENGTH_GENOTYPE_STRING_TESTS) {
//
//
// TESTING MULTIPLE SIZED LISTS IN THE GENOTYPE FIELD
//
//
addGenotypeTests(
site,
attr("g1", ref, "GS", Arrays.asList("S1", "S2")),
attr("g2", ref, "GS", Arrays.asList("S3", "S4")));
addGenotypeTests(
site, // g1 is missing the string, and g2 is missing FLOAT1
attr("g1", ref, "FLOAT1", 1.0),
attr("g2", ref, "GS", Arrays.asList("S3", "S4")));
// variable sized lists
addGenotypeTests(
site, attr("g1", ref, "GV", "S1"), attr("g2", ref, "GV", Arrays.asList("S3", "S4")));
addGenotypeTests(
site,
attr("g1", ref, "GV", Arrays.asList("S1", "S2")),
attr("g2", ref, "GV", Arrays.asList("S3", "S4", "S5")));
addGenotypeTests(
site, // missing value in varlist of string
attr("g1", ref, "FLOAT1", 1.0),
attr("g2", ref, "GV", Arrays.asList("S3", "S4", "S5")));
}
//
//
// TESTING GENOTYPE FILTERS
//
//
addGenotypeTests(
site,
new GenotypeBuilder("g1-x", Arrays.asList(ref, ref)).filters("X").make(),
new GenotypeBuilder("g2-x", Arrays.asList(ref, ref)).filters("X").make());
addGenotypeTests(
site,
new GenotypeBuilder("g1-unft", Arrays.asList(ref, ref)).unfiltered().make(),
new GenotypeBuilder("g2-x", Arrays.asList(ref, ref)).filters("X").make());
addGenotypeTests(
site,
new GenotypeBuilder("g1-unft", Arrays.asList(ref, ref)).unfiltered().make(),
new GenotypeBuilder("g2-xy", Arrays.asList(ref, ref)).filters("X", "Y").make());
addGenotypeTests(
site,
new GenotypeBuilder("g1-unft", Arrays.asList(ref, ref)).unfiltered().make(),
new GenotypeBuilder("g2-x", Arrays.asList(ref, ref)).filters("X").make(),
new GenotypeBuilder("g3-xy", Arrays.asList(ref, ref)).filters("X", "Y").make());
}
/**
* Read in a list of ExactCall objects from reader, keeping only those with starts in startsToKeep
* or all sites (if this is empty)
*
* @param reader a just-opened reader sitting at the start of the file
* @param startsToKeep a list of start position of the calls to keep, or empty if all calls should
* be kept
* @param parser a genome loc parser to create genome locs
* @return a list of ExactCall objects in reader
* @throws IOException
*/
public static List<ExactCall> readExactLog(
final BufferedReader reader, final List<Integer> startsToKeep, GenomeLocParser parser)
throws IOException {
if (reader == null) throw new IllegalArgumentException("reader cannot be null");
if (startsToKeep == null) throw new IllegalArgumentException("startsToKeep cannot be null");
if (parser == null) throw new IllegalArgumentException("GenomeLocParser cannot be null");
List<ExactCall> calls = new LinkedList<ExactCall>();
// skip the header line
reader.readLine();
// skip the first "type" line
reader.readLine();
while (true) {
final VariantContextBuilder builder = new VariantContextBuilder();
final List<Allele> alleles = new ArrayList<Allele>();
final List<Genotype> genotypes = new ArrayList<Genotype>();
final double[] posteriors = new double[2];
final double[] priors = MathUtils.normalizeFromLog10(new double[] {0.5, 0.5}, true);
final List<Integer> mle = new ArrayList<Integer>();
final Map<Allele, Double> log10pNonRefByAllele = new HashMap<Allele, Double>();
long runtimeNano = -1;
GenomeLoc currentLoc = null;
while (true) {
final String line = reader.readLine();
if (line == null) return calls;
final String[] parts = line.split("\t");
final GenomeLoc lineLoc = parser.parseGenomeLoc(parts[0]);
final String variable = parts[1];
final String key = parts[2];
final String value = parts[3];
if (currentLoc == null) currentLoc = lineLoc;
if (variable.equals("type")) {
if (startsToKeep.isEmpty() || startsToKeep.contains(currentLoc.getStart())) {
builder.alleles(alleles);
final int stop = currentLoc.getStart() + alleles.get(0).length() - 1;
builder.chr(currentLoc.getContig()).start(currentLoc.getStart()).stop(stop);
builder.genotypes(genotypes);
final int[] mleInts = ArrayUtils.toPrimitive(mle.toArray(new Integer[] {}));
final AFCalcResult result =
new AFCalcResult(mleInts, 1, alleles, posteriors, priors, log10pNonRefByAllele);
calls.add(new ExactCall(builder.make(), runtimeNano, result));
}
break;
} else if (variable.equals("allele")) {
final boolean isRef = key.equals("0");
alleles.add(Allele.create(value, isRef));
} else if (variable.equals("PL")) {
final GenotypeBuilder gb = new GenotypeBuilder(key);
gb.PL(GenotypeLikelihoods.fromPLField(value).getAsPLs());
genotypes.add(gb.make());
} else if (variable.equals("log10PosteriorOfAFEq0")) {
posteriors[0] = Double.valueOf(value);
} else if (variable.equals("log10PosteriorOfAFGt0")) {
posteriors[1] = Double.valueOf(value);
} else if (variable.equals("MLE")) {
mle.add(Integer.valueOf(value));
} else if (variable.equals("pNonRefByAllele")) {
final Allele a = Allele.create(key);
log10pNonRefByAllele.put(a, Double.valueOf(value));
} else if (variable.equals("runtime.nano")) {
runtimeNano = Long.valueOf(value);
} else {
// nothing to do
}
}
}
}
private Genotype makeG(String sample, Allele a1, Allele a2) {
return GenotypeBuilder.create(sample, Arrays.asList(a1, a2));
}