[InlineData("C", ".", "DP:AU:CU:GU:TU", "75:0,0:72,77:0,0:0,2")] // ref minor (AC)
public void VariantFrequency_ReturnNull(string refAllele, string altAllele, string formatCol, string sampleCol)
{
var vcfLine = $"chr1\t5592503\t.\t{refAllele}\t{altAllele}\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.Null(sample.VariantFrequencies);
}
public void EmptySample()
{
const string vcfLine = "chr7 127717248 MantaINV:267944:0:1:2:0:0 T <INV> . PASS END=140789466;SVTYPE=INV;SVLEN=13072218;INV5 PR:SR .";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.True(sample.IsEmpty);
}
public void DeNovoQuality()
{
const string vcfLine = "chr1\t5592503\t.\tC\tT\t900.00\tPASS\t.\tGT:DQ\t0/1:20";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.Equal(20, sample.DeNovoQuality);
}
public void PiscesTotalDepth()
{
const string vcfLine =
"chr1\t115251293\t.\tGA\tG\t100\tSB;LowVariantFreq\tDP=7882\tGT:GQ:AD:VF:NL:SB:GQX\t0/1:100:7588,294:0:20:-100.0000:100";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns, 7882);
var samples = extractor.ExtractSamples();
var sample = samples[0];
var observedTotalDepth = sample.TotalDepth;
const int expectedTotalDepth = 7882;
Assert.Equal(expectedTotalDepth, observedTotalDepth);
}
public void TotalDepth(string formatCol, string sampleCol, int?expectedTotalDepth)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
var observedTotalDepth = sample?.TotalDepth;
Assert.Equal(expectedTotalDepth, observedTotalDepth);
}
public void Genotype(string sampleCol, string expectedGenotype)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT\t900.00\tPASS\t.\tGT:GQ:GQX:DP:DPF:AD\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
var observedGenotype = sample?.Genotype;
Assert.Equal(expectedGenotype, observedGenotype);
}
public void FailedFilter(string sampleCol, bool?expectedFailedFilter)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT\t900.00\tPASS\t.\tGT:GQ:GQX:DP:DPF:FT\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
var observedFailedFilter = sample?.FailedFilter;
Assert.Equal(expectedFailedFilter, observedFailedFilter);
}
public void VariantFrequencyMultiAllelic(string formatCol, string sampleCol, string expectedVariantFrequency)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT,A\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(1, samples.Count);
var sample = samples[0];
var observedVariantFrequency = sample?.VariantFrequency;
Assert.Equal(expectedVariantFrequency, observedVariantFrequency);
}
public void AlleleDepthsMultiAllelic(string formatCol, string sampleCol, int[] expectedAlleleDepths)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT,A\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
var observedAlleleDepths = sample?.AlleleDepths;
Assert.Equal(expectedAlleleDepths, observedAlleleDepths);
}
public void GenotypeQuality(string formatCol, string sampleCol, string expectedGenotypeQuality)
{
string vcfLine = $"chr1\t5592503\t.\tC\tT\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(1, samples.Count);
var sample = samples[0];
var observedGenotypeQuality = sample?.GenotypeQuality;
Assert.Equal(expectedGenotypeQuality, observedGenotypeQuality);
}
public void ArtifactAdjustedQualityScore_LikelihoodRatioQualityScore()
{
const string vcfLine = "chr1\t2488109\t.\tG\tA\t5\tLowSupport\tDP=339\tGT:GQ:AD:DP:VF:NL:SB:NC:US:AQ:LQ\t0/1:5:338,1:339:0.00295:30:-7.3191:0.0314:0,0,0,1,0,0,17,1,129,21,148,22:3.366:0.001";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.NotNull(sample.ArtifactAdjustedQualityScore);
Assert.NotNull(sample.LikelihoodRatioQualityScore);
Assert.Equal("3.366", sample.ArtifactAdjustedQualityScore.Value.ToString("0.###"));
Assert.Equal("0.001", sample.LikelihoodRatioQualityScore.Value.ToString("0.###"));
}
[InlineData("T", "GT:AD:DP:VF", "0/1:317,200:517:0.38685", "0.3869")] // VF (rounding issue)
public void VariantFrequency_Nominal(string altAllele, string formatCol, string sampleCol, string expectedResults)
{
string vcfLine = $"chr1\t5592503\t.\tC\t{altAllele}\t900.00\tPASS\t.\t{formatCol}\t{sampleCol}";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.NotNull(sample?.VariantFrequencies);
var observedResults = string.Join(',', sample.VariantFrequencies.Select(x => x.ToString("0.####")));
Assert.Equal(expectedResults, observedResults);
}
public void EmptySamples()
{
// for NIR-1306
const string vcfLine = "chrX 2735147 . G A 38.25 VQSRTrancheSNP99.90to100.00 AC=3;AF=0.500;AN=6;BaseQRankSum=-0.602;DP=56;Dels=0.00;FS=30.019;HaplotypeScore=7.7259;MLEAC=3;MLEAF=0.500;MQ=41.18;MQ0=0;MQRankSum=0.098;QD=1.06;ReadPosRankSum=0.266;SB=-8.681e-03;VQSLOD=-6.0901;culprit=QD GT:AD:DP:GQ:PL 0:7,0:7:3:0,3,39 ./. 0/1:14,3:17:35:35,0,35 1/1:9,10:19:3:41,3,0";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(4, samples.Length);
var sample = samples[1];
var observedGenotype = sample.Genotype;
var observedVariantFrequency = sample.VariantFrequencies;
Assert.Equal("./.", observedGenotype);
Assert.Null(observedVariantFrequency);
}
public void MajorChromosomeCopyTest()
{
// data from NIR-1095
// for NIR-1218
const string vcfLine = "1 9314202 Canvas:GAIN:1:9314202:9404148 N <CNV> 36 PASS SVTYPE=CNV;END=9404148;ensembl_gene_id=ENSG00000049239,ENSG00000252841,ENSG00000171621 RC:BC:CN:MCC . 151:108:6:4";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(2, samples.Length);
var sample = samples[1];
var observedMcc = sample?.IsLossOfHeterozygosity;
Assert.False(observedMcc);
}
public void Smn1()
{
const string vcfLine = "5\t70247773\t.\tC\tT\t366\tPASS\tSNVHPOL=4;MQ=60\tGT:DST:DID:DCS:SCH:PCN:PLG:MAD:GQ:GQX:DP:DPF:AD:ADF:ADR:SB:FT:PL\t0/1:-:70:Orphanet:-:3,3:6606,6607:41,49:368:364:81:11:39,42:21,20:18,22:-41.0:PASS:370,0,365";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Single(samples);
var sample = samples[0];
Assert.Equal(new[] { "-" }, sample?.DiseaseAffectedStatus);
Assert.Equal(new[] { "70" }, sample?.DiseaseIds);
Assert.Equal(new[] { "Orphanet" }, sample?.DiseaseClassificationSources);
Assert.Equal("-", sample?.SilentCarrierHaplotype);
Assert.Equal(new[] { 3, 3 }, sample?.ParalogousGeneCopyNumbers);
Assert.Equal(new[] { 6606, 6607 }, sample?.ParalogousEntrezGeneIds);
Assert.Equal(new[] { 41, 49 }, sample?.MpileupAlleleDepths);
}
public void SplitReadCounts()
{
const string vcfLine = "chr7 127717248 MantaINV:267944:0:1:2:0:0 T <INV> . PASS END=140789466;SVTYPE=INV;SVLEN=13072218;INV5 PR:SR 78,0:65,0 157,42:252,63";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(2, samples.Length);
var sample1 = samples[0];
Assert.Equal(new[] { 78, 0 }, sample1.PairEndReadCounts);
Assert.Equal(new[] { 65, 0 }, sample1.SplitReadCounts);
var sample2 = samples[1];
Assert.Equal(new[] { 157, 42 }, sample2.PairEndReadCounts);
Assert.Equal(new[] { 252, 63 }, sample2.SplitReadCounts);
}
public void SplitReadCounts()
{
var vcfLine =
"chr7 127717248 MantaINV:267944:0:1:2:0:0 T <INV> . PASS END=140789466;SVTYPE=INV;SVLEN=13072218;INV5 PR:SR 78,0:65,0 157,42:252,63";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples();
Assert.Equal(2, samples.Count);
var sample1 = samples[0];
Assert.Equal(new[] { "78", "0" }, sample1.PairEndReadCounts);
Assert.Equal(new[] { "65", "0" }, sample1.SplitReadCounts);
var sample2 = samples[1];
Assert.Equal(new[] { "157", "42" }, sample2.PairEndReadCounts);
Assert.Equal(new[] { "252", "63" }, sample2.SplitReadCounts);
}
public void GatkGenomeVcfSample()
{
// For NIR-1320
// the original AD= 15,11,0
var vcfLine =
"1 30923 rs140337953 G T,<NON_REF> 264.77 PASS BaseQRankSum=0.259;DB;DP=26;MLEAC=1,0;MLEAF=0.500,0.00;MQ=43.87;MQ0=0;MQRankSum=-0.830;ReadPosRankSum=-0.156 GT:AD:GQ:PL:SB 0/1:15,11,20:99:293,0,330,337,363,700:8,7,3,8";
var vcfColumns = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(vcfColumns);
var samples = extractor.ExtractSamples(true);
Assert.Equal(1, samples.Count);
var sample = samples[0];
var observedVariantFrequency = sample?.VariantFrequency;
var expectedVariantFrequency = "0.4231";
Assert.Equal(expectedVariantFrequency, observedVariantFrequency);
var expectedAlleleDepths = new[] { "15", "11" };
var observedAlleleDepths = sample?.AlleleDepths;
Assert.Equal(expectedAlleleDepths, observedAlleleDepths);
}
public void OutputCanvasCnvRelevantField()
{
var vcfLine = "1 9314201 Canvas:GAIN:1:9314202:9404148 N <CNV> 36 PASS SVTYPE=CNV;END=9404148;ensembl_gene_id=ENSG00000049239,ENSG00000252841,ENSG00000171621 RC:BC:CN:MCC . 151:108:6:4";
var annotatedVariant = DataUtilities.GetVariant(Resources.CacheGRCh37("ENST00000377403_chr1_Ensembl84"), null, vcfLine);
Assert.NotNull(annotatedVariant);
JsonUtilities.AlleleEquals(annotatedVariant,
"{\"altAllele\":\"CNV\",\"refAllele\":\"N\",\"begin\":9314202,\"chromosome\":\"1\",\"end\":9404148,\"variantType\":\"copy_number_variation\",\"vid\":\"1:9314202:9404148:6\",\"overlappingGenes\":[\"H6PD\"],\"transcripts\":{\"ensembl\":[{\"transcript\":\"ENST00000377403.2\",\"bioType\":\"protein_coding\",\"exons\":\"4-5/5\",\"introns\":\"3-4/4\",\"geneId\":\"ENSG00000049239\",\"hgnc\":\"H6PD\",\"consequence\":[\"copy_number_increase\"],\"isCanonical\":true,\"proteinId\":\"ENSP00000366620.1\"}]}}");
var cols = vcfLine.Split('\t');
var extractor = new SampleFieldExtractor(cols);
var samples = extractor.ExtractSamples();
Assert.Equal(2, samples.Count);
var sample = samples[1];
var observedCn = sample?.CopyNumber;
Assert.Equal("6", observedCn);
}