public void VariantGenerator_ForceGenotype_ConsistentAllele_Recompose()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAATCGCGA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 1|1 0/0", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C G . PASS . GT 0|1 0|1", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 4 . C G,A . PASS . GT 0|1 0/0", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 6
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGC TGC,TGG . PASS RECOMPOSED GT 1|2 .", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_MinGQUsed_DotAndNullIgnored()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:PS:GQ 0|1:123:. 2/2:.:14.2 0|2:456:.", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G . PASS . GT:PS:GQ 1|1:301:. 1|2:.:18 1|2:456:15.6", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT . 1|0 0/1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Equal(2, recomposedPositions.Count);
Assert.Equal("chr1 2 . AGC AGA,GGG,TGA . PASS RECOMPOSED GT:GQ:PS 1|3:.:123 . 1|2:15.6:456", string.Join("\t", recomposedPositions[0].VcfFields));
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG . PASS RECOMPOSED GT:GQ . 1|2:14.2 .", string.Join("\t", recomposedPositions[1].VcfFields));
}
public void VariantGenerator_FilterTag_DotTreatedAsPass()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:PS 0|1:123 2/2:. 0|2:456", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G . . . GT:PS 1|1:301 1|2:. 1|2:456", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . FailedForSomeReason . GT:PS . 1|0:. 0/1:456", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Equal(2, recomposedPositions.Count);
Assert.Equal("chr1 2 . AGC AGA,GGG,TGA . PASS RECOMPOSED GT:PS 1|3:123 . 1|2:456", string.Join("\t", recomposedPositions[0].VcfFields));
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG . FilteredVariantsRecomposed RECOMPOSED GT . 1|2 .", string.Join("\t", recomposedPositions[1].VcfFields));
}
public void VariantGenerator_FilterTag_PassedMnvOverridesFailedOne()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 0|1 0|1", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A . FailedForSomeReason . GT 0|0 0|1", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT 0|1 0|1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
6, 8, 10
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGCTG TGATC,TGCTC . PASS RECOMPOSED GT 0|2 0|1", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_HomoReferenceGenotype_Output()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 1 . C A . PASS . GT:PS 0|1:1584593 0/0:.", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 2 . A C . PASS . GT:PS 0|1:1584593 0/0:.", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
3, 4
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 1 . CA AC . PASS RECOMPOSED GT:PS 0|1:1584593 0|0", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_OverlappingDeletionAtTheEnd_Ignored()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAATCGCGA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 0|1 0/0", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A . PASS . GT 1|1 0/0", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 4 . CTGAATCGCGA C . PASS . GT 0/0 0|1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 6
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGC AGA,TGA . PASS RECOMPOSED GT 1|2 0|0", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_MinQualUsed_DotIgnored()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:PS 0|1:123 2/2:. 0|2:456", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G 45 PASS . GT:PS 1|1:301 1|2:. 1|2:456", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C 30.1 PASS . GT . 1|0 0/1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Equal(2, recomposedPositions.Count);
Assert.Equal("chr1 2 . AGC AGA,GGG,TGA 45 PASS RECOMPOSED GT:PS 1|3:123 . 1|2:456", string.Join("\t", recomposedPositions[0].VcfFields));
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG 30.1 PASS RECOMPOSED GT . 1|2 .", string.Join("\t", recomposedPositions[1].VcfFields));
}
public void ParseVcfLine_line_with_only_NonRef_is_refMinor()
{
const string vcfLine = "1 10628385 . C <NON_REF> . LowGQX;HighDPFRatio END=10628385;BLOCKAVG_min30p3a GT:GQX:DP:DPF 0/0:24:9:18";
var chromosome = new Chromosome("chr1", "1", 0);
var refMinorProvider = new Mock <IRefMinorProvider>();
refMinorProvider.Setup(x => x.GetGlobalMajorAllele(chromosome, 10628385)).Returns("T");
var seqProvider = ParserTestUtils.GetSequenceProvider(10628385, "C", 'A',
new Dictionary <string, IChromosome> {
["1"] = chromosome
});
var refNameToChromosome = seqProvider.RefNameToChromosome;
var variantFactory = new VariantFactory(seqProvider);
var position = AnnotationUtilities.ParseVcfLine(vcfLine, refMinorProvider.Object, variantFactory, refNameToChromosome);
var annotatedVariants = Annotator.GetAnnotatedVariants(position.Variants);
Assert.Equal("C", position.RefAllele);
Assert.Equal(new[] { "<NON_REF>" }, position.AltAlleles);
Assert.Equal("T", position.Variants[0].RefAllele);
Assert.Equal("C", position.Variants[0].AltAllele);
// Variants
Assert.Equal(new[] { "C" }, annotatedVariants.Select(x => x.Variant.AltAllele).ToArray());
}
public void VariantGenerator_AllTrailingMissingValuesDroped()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:GQ:PS 0|1:.:123 2/2 1|1:17:456", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G . PASS . GT:GQ:PS ./. 1|2 1|2:15.6:456", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT:GQ:PS ./. 1|0 1|1:13:456", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG,TGATC,TGGTC . PASS RECOMPOSED GT:GQ:PS . 1|2 3|4:13:456", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_RefAllelesAddedToMergeAlleles()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAACT"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 0|1 0|0 0|1 0|0", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A . PASS . GT 1|1 1|1 1|1 1|1", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT 1|1 1|1 1|1 1|1", sequenceProvider.RefNameToChromosome);
var position4 = AnnotationUtilities.GetSimplePosition("chr1 8 . A G . PASS . GT 0|1 0|1 0|0 0|0", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8, 10
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3, position4
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGCTGAA AGATCAA,AGATCAG,TGATCAA,TGATCAG . PASS RECOMPOSED GT 1|4 1|2 1|3 1|1", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_HomozygousSitesAndPhasedSites_Recomposed()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 1/1 1/1 1/1", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 3 . G A,G 45 PASS . GT:PS 1|1:2 1|2:2 1|2:2", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A 45 PASS . GT 1/1 1/1 1/1", sequenceProvider.RefNameToChromosome);
var position4 = AnnotationUtilities.GetSimplePosition("chr1 5 . T A,G 45 PASS . GT:PS 1|1:4 1|2:2 1|2:4", sequenceProvider.RefNameToChromosome);
var position5 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C 30 PASS . GT 1/1 1/1 1/1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 5, 6, 7, 8
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3, position4, position5
}, functionBlockRanges).ToList();
Assert.Equal(3, recomposedPositions.Count);
Assert.Equal("chr1 2 . AGC TAA,TGA 45 PASS RECOMPOSED GT:PS . . 1|2:2", string.Join("\t", recomposedPositions[0].VcfFields));
Assert.Equal("chr1 2 . AGCTG TAAAC,TGAGC 30 PASS RECOMPOSED GT:PS 1|1 1|2:2 .", string.Join("\t", recomposedPositions[1].VcfFields));
Assert.Equal("chr1 4 . CTG AAC,AGC 30 PASS RECOMPOSED GT:PS . . 1|2:4", string.Join("\t", recomposedPositions[2].VcfFields));
}
public void VariantGenerator_ConflictAltAlleles_AlleleBlockStartInTheMiddle_NoRecompositionNoException()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAATCGCGA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 1 . C T . PASS . GT 0/1 0|1", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 0/1 0|0", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 4 . C G . PASS . GT 1|1 1|1", sequenceProvider.RefNameToChromosome);
var position4 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A . PASS . GT 1|1 0|1", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
3, 4, 6, 6
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3, position4
}, functionBlockRanges).ToList();
Assert.Empty(recomposedPositions);
}
public void GenerateOutput_NothingRecomposed_ReturnOriginalVcfFieldList()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var variantGenerator = new VariantGenerator(sequenceProvider);
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT:PS 0|1:123", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C G . PASS . GT 0/1", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT 0|1", sequenceProvider.RefNameToChromosome);
var positions = new List <ISimplePosition> {
position1, position2, position3
};
var recomposable = new List <bool> {
true, true, true
};
var functionBlockRanges = new List <int> {
4, 6, 8
};
var bufferedPositions = new BufferedPositions(positions, recomposable, functionBlockRanges);
var positionProcessor = new PositionProcessor(_positionBufferMock.Object, variantGenerator);
var output = positionProcessor.GenerateOutput(bufferedPositions).ToArray();
for (int i = 0; i < output.Length; i++)
{
Assert.True(positions[i].VcfFields.SequenceEqual(output[i].VcfFields));
}
}
public void ParseVcfLine_NonInformativeAlleles_WithNormalAllele_NotFiltered()
{
const string vcfLine1 = "chr1 13133 . T <*>,G 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
const string vcfLine2 = "chr1 13133 . T *,C 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
const string vcfLine3 = "chr1 13133 . T <M>,A 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
const string vcfLine4 = "chr1 13133 . T A,<NON_REF> 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(13133, "T", 'A', ChromosomeUtilities.RefNameToChromosome);
var refNameToChromosome = seqProvider.RefNameToChromosome;
var variantFactory = new VariantFactory(seqProvider);
var position1 = AnnotationUtilities.ParseVcfLine(vcfLine1, refMinorProvider.Object, variantFactory, refNameToChromosome);
var position2 = AnnotationUtilities.ParseVcfLine(vcfLine2, refMinorProvider.Object, variantFactory, refNameToChromosome);
var position3 = AnnotationUtilities.ParseVcfLine(vcfLine3, refMinorProvider.Object, variantFactory, refNameToChromosome);
var position4 = AnnotationUtilities.ParseVcfLine(vcfLine4, refMinorProvider.Object, variantFactory, refNameToChromosome);
var annotatedVariants1 = Annotator.GetAnnotatedVariants(position1.Variants);
var annotatedVariants2 = Annotator.GetAnnotatedVariants(position2.Variants);
var annotatedVariants3 = Annotator.GetAnnotatedVariants(position3.Variants);
var annotatedVariants4 = Annotator.GetAnnotatedVariants(position4.Variants);
// SimplePositions
Assert.Equal(new[] { "<*>", "G" }, position1.AltAlleles);
Assert.Equal(new[] { "*", "C" }, position2.AltAlleles);
Assert.Equal(new[] { "<M>", "A" }, position3.AltAlleles);
Assert.Equal(new[] { "A", "<NON_REF>" }, position4.AltAlleles);
// Variants
Assert.Equal(new[] { "<*>", "G" }, annotatedVariants1.Select(x => x.Variant.AltAllele).ToArray());
Assert.Equal(new[] { "*", "C" }, annotatedVariants2.Select(x => x.Variant.AltAllele).ToArray());
Assert.Equal(new[] { "<M>", "A" }, annotatedVariants3.Select(x => x.Variant.AltAllele).ToArray());
Assert.Equal(new[] { "A", "<NON_REF>" }, annotatedVariants4.Select(x => x.Variant.AltAllele).ToArray());
}
public void ParseVcfLine_NonInformativeAlleles_Alone_NotFiltered()
{
const string vcfLine1 = "chr1 13133 . T <*> 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
const string vcfLine2 = "chr1 13133 . T * 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
const string vcfLine3 = "chr1 13133 . T <M> 36.00 PASS SNVSB=0.0;SNVHPOL=4 GT:GQ:GQX:DP:DPF:AD 0/1:62:20:7:1:3,4";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(13133, "T", 'A', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider);
var position1 = AnnotationUtilities.ParseVcfLine(vcfLine1, refMinorProvider.Object, variantFactory, seqProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.ParseVcfLine(vcfLine2, refMinorProvider.Object, variantFactory, seqProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.ParseVcfLine(vcfLine3, refMinorProvider.Object, variantFactory, seqProvider.RefNameToChromosome);
var annotatedVariants1 = Annotator.GetAnnotatedVariants(position1.Variants);
var annotatedVariants2 = Annotator.GetAnnotatedVariants(position2.Variants);
var annotatedVariants3 = Annotator.GetAnnotatedVariants(position3.Variants);
// SimplePositions unchanged
Assert.Equal("<*>", position1.AltAlleles[0]);
Assert.Equal("*", position2.AltAlleles[0]);
Assert.Equal("<M>", position3.AltAlleles[0]);
// Variants not filtered
Assert.Equal("<*>", annotatedVariants1[0].Variant.AltAllele);
Assert.Equal("*", annotatedVariants2[0].Variant.AltAllele);
Assert.Equal("<M>", annotatedVariants3[0].Variant.AltAllele);
}
public void ParseVcfLine_line_with_only_NonRef_is_not_refMinor()
{
const string vcfLine = "1 10005 . C <NON_REF> . LowGQX END=10034;BLOCKAVG_min30p3a GT:GQX:DP:DPF 0/0:3:1:0";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider =
ParserTestUtils.GetSequenceProvider(10005, "C", 'A', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider.Sequence, _vidCreator);
var position = AnnotationUtilities.ParseVcfLine(vcfLine, refMinorProvider.Object, seqProvider, null, variantFactory);
var annotatedVariants = Annotator.GetAnnotatedVariants(position.Variants);
Assert.Equal("C", position.RefAllele);
Assert.Equal(new[] { "<NON_REF>" }, position.AltAlleles);
Assert.Null(position.Variants);
Assert.Null(annotatedVariants);
}
public void GetJsonString_fisherStrand()
{
const string vcfLine = "21\t9411410\t.\tC\tT\t9.51\tDRAGENSnpHardQUAL\tAC=2;AF=1.000;AN=2;DP=2;FS=0.000;MQ=100.00;QD=9.51;SOR=1.609";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(9411410, "C", 'A', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider.Sequence, new VariantId());
var position = AnnotationUtilities.ParseVcfLine(vcfLine, refMinorProvider.Object, seqProvider, null, variantFactory);
IVariant[] variants = GetVariants();
IAnnotatedVariant[] annotatedVariants = Annotator.GetAnnotatedVariants(variants);
var annotatedPosition = new AnnotatedPosition(position, annotatedVariants);
string observedResult = annotatedPosition.GetJsonString();
Assert.NotNull(observedResult);
Assert.Contains("\"fisherStrandBias\":0", observedResult);
}
public void Test_crash_caused_by_variant_trimming()
{
const string vcfLine1 = "chr1 8021910 rs373653682 GGTGCTGGACGGTGTCCCT G . . .";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(8021910, "GGTGCTGGACGGTGTCCCT", 'A', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider.Sequence, _vidCreator);
var position1 = AnnotationUtilities.ParseVcfLine(vcfLine1, refMinorProvider.Object, seqProvider, null, variantFactory);
var annotatedVariants1 = Annotator.GetAnnotatedVariants(position1.Variants);
// SimplePositions
Assert.Equal(new[] { "G" }, position1.AltAlleles);
// Variants
Assert.Equal(new[] { "" }, annotatedVariants1.Select(x => x.Variant.AltAllele).ToArray());
}
public void GetJsonString_BreakEndEventId()
{
const string vcfLine = "1\t38432782\tMantaBND:2312:0:1:0:0:0:0\tG\tG]6:28863899]\t971\tPASS\tSVTYPE=BND;MATEID=MantaBND:2312:0:1:0:0:0:1;EVENT=MantaBND:2312:0:1:0:0:0:0;JUNCTION_QUAL=716;BND_DEPTH=52;MATE_BND_DEPTH=56";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(38432782, "G", 'C', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider.Sequence, new VariantId());
var position = AnnotationUtilities.ParseVcfLine(vcfLine, refMinorProvider.Object, seqProvider, null, variantFactory);
IVariant[] variants = GetVariants();
IAnnotatedVariant[] annotatedVariants = Annotator.GetAnnotatedVariants(variants);
var annotatedPosition = new AnnotatedPosition(position, annotatedVariants);
string observedResult = annotatedPosition.GetJsonString();
Assert.NotNull(observedResult);
Assert.Contains("\"breakendEventId\":\"MantaBND:2312:0:1:0:0:0:0\"", observedResult);
}
public void GenerateOutput_Return_OriginalAndRecomposed_VcfFieldList()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var variantGenerator = new VariantGenerator(sequenceProvider);
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT:PS 0|1:.", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C G . PASS . GT 1/1", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT 0|1", sequenceProvider.RefNameToChromosome);
var positions = new List <ISimplePosition> {
position1, position2, position3
};
var recomposable = new List <bool> {
true, true, true
};
var functionBlockRanges = new List <int> {
4, 6, 8
};
var bufferedPositions = new BufferedPositions(positions, recomposable, functionBlockRanges);
var positionProcessor = new PositionProcessor(_positionBufferMock.Object, variantGenerator);
var output = positionProcessor.GenerateOutput(bufferedPositions).ToArray();
var expectedOutput = new string[4][];
expectedOutput[0] = position1.VcfFields;
expectedOutput[1] = new[]
{ "chr1", "2", ".", "AGCTG", "AGGTG,TGGTC", ".", "PASS", "RECOMPOSED", "GT", "1|2" };
expectedOutput[2] = position2.VcfFields;
expectedOutput[3] = position3.VcfFields;
for (int i = 0; i < output.Length; i++)
{
Assert.True(expectedOutput[i].SequenceEqual(output[i].VcfFields));
}
}
public void GetJsonString_StrelkaSomatic()
{
const string vcfLine = "chr1 13813 . T G . LowQscore SOMATIC;QSS=33;TQSS=1;NT=ref;QSS_NT=16;TQSS_NT=1;SGT=TT->GT;DP=266;MQ=23.89;MQ0=59;ALTPOS=69;ALTMAP=37;ReadPosRankSum=1.22;SNVSB=5.92;PNOISE=0.00;PNOISE2=0.00;VQSR=1.93";
var refMinorProvider = new Mock <IRefMinorProvider>();
var seqProvider = ParserTestUtils.GetSequenceProvider(13813, "T", 'C', ChromosomeUtilities.RefNameToChromosome);
var variantFactory = new VariantFactory(seqProvider.Sequence, new VariantId());
var position = AnnotationUtilities.ParseVcfLine(vcfLine, refMinorProvider.Object, seqProvider, variantFactory);
IVariant[] variants = GetVariants();
IAnnotatedVariant[] annotatedVariants = Annotator.GetAnnotatedVariants(variants);
var annotatedPosition = new AnnotatedPosition(position, annotatedVariants);
string observedResult = annotatedPosition.GetJsonString();
Assert.NotNull(observedResult);
Assert.Contains("\"jointSomaticNormalQuality\":16", observedResult);
Assert.Contains("\"recalibratedQuality\":1.93", observedResult);
}
public void VariantGenerator_NoMnvAfterTrimming_NotRecompose()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 1 . C A . PASS . GT:PS 1|0:1584593 1|1:. 0|1:.", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 2 . A C . PASS . GT:PS 0|1:1584593 0/0:. 0/0:.", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
3, 4
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2
}, functionBlockRanges).ToList();
Assert.Empty(recomposedPositions);
}
public void VariantGenerator_SampleColumnCorrectlyProcessed_WhenTrailingMissingValuesDroped()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:PS:GQ 0|1:123 2/2:.:14.2 ./.", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G . PASS . GT:PS:GQ ./. 1|2:.:18 1|2:456:15.6", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT ./. 1|0 ./.", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG . PASS RECOMPOSED GT:GQ . 1|2:14.2 .", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_FilterTag_OnlyDecomposedVariantsConsidered()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(ChromosomeUtilities.RefNameToChromosome);
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T . PASS . GT 0|1 0/1 0|0", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A . FailedForSomeReason . GT 0|0 0/1 0|0", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT 1|1 0/1 0|0", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
6, 8, 10
};
var recomposer = new VariantGenerator(sequenceProvider, _vidCreator);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Single(recomposedPositions);
Assert.Equal("chr1 2 . AGCTG AGCTC,TGCTC . PASS RECOMPOSED GT 1|2 . 0|0", string.Join("\t", recomposedPositions[0].VcfFields));
}
public void VariantGenerator_AsExpected()
{
var mockSequenceProvider = new Mock <ISequenceProvider>();
mockSequenceProvider.SetupGet(x => x.RefNameToChromosome)
.Returns(new Dictionary <string, IChromosome> {
{ "chr1", new Chromosome("chr1", "1", 0) }
});
mockSequenceProvider.SetupGet(x => x.Sequence).Returns(new SimpleSequence("CAGCTGAA"));
var sequenceProvider = mockSequenceProvider.Object;
var position1 = AnnotationUtilities.GetSimplePosition("chr1 2 . A T,G . PASS . GT:PS 0|1:123 2/2:789 0|2:456", sequenceProvider.RefNameToChromosome);
var position2 = AnnotationUtilities.GetSimplePosition("chr1 4 . C A,G . PASS . GT:PS 1|1:301 1|2:789 1|2:456", sequenceProvider.RefNameToChromosome);
var position3 = AnnotationUtilities.GetSimplePosition("chr1 6 . G C . PASS . GT:PS . 1|0:789 0/1:.", sequenceProvider.RefNameToChromosome);
var functionBlockRanges = new List <int> {
4, 6, 8
};
var recomposer = new VariantGenerator(sequenceProvider);
var recomposedPositions = recomposer.Recompose(new List <ISimplePosition> {
position1, position2, position3
}, functionBlockRanges).ToList();
Assert.Equal(2, recomposedPositions.Count);
Assert.Equal("chr1 2 . AGC AGA,GGG,TGA . PASS RECOMPOSED GT:PS 1|3:123 . 1|2:456", string.Join("\t", recomposedPositions[0].VcfFields));
Assert.Equal("chr1 2 . AGCTG GGATC,GGGTG . PASS RECOMPOSED GT:PS . 1|2:789 .", string.Join("\t", recomposedPositions[1].VcfFields));
//Check LinkedVids
//SNVs
Assert.Equal(2, position1.LinkedVids.Length);
Assert.Equal(new List <string> {
"1:2:4:TGA"
}, position1.LinkedVids[0]);
position1.LinkedVids[1].Sort();
Assert.Equal(new List <string> {
"1:2:4:GGG", "1:2:6:GGATC", "1:2:6:GGGTG"
}, position1.LinkedVids[1]);
Assert.Equal(2, position2.LinkedVids.Length);
position2.LinkedVids[0].Sort();
Assert.Equal(new List <string> {
"1:2:4:AGA", "1:2:4:TGA", "1:2:6:GGATC"
}, position2.LinkedVids[0]);
position2.LinkedVids[1].Sort();
Assert.Equal(new List <string> {
"1:2:4:GGG", "1:2:6:GGGTG"
}, position2.LinkedVids[1]);
Assert.Single(position3.LinkedVids);
Assert.Equal(new List <string> {
"1:2:6:GGATC"
}, position3.LinkedVids[0]);
//MNVs
Assert.Equal(3, recomposedPositions[0].LinkedVids.Length);
Assert.Equal(new List <string> {
"1:4:A"
}, recomposedPositions[0].LinkedVids[0]);
Assert.Equal(new List <string> {
"1:2:G", "1:4:G"
}, recomposedPositions[0].LinkedVids[1]);
Assert.Equal(new List <string> {
"1:2:T", "1:4:A"
}, recomposedPositions[0].LinkedVids[2]);
Assert.Equal(2, recomposedPositions[1].LinkedVids.Length);
Assert.Equal(new List <string> {
"1:2:G", "1:4:A", "1:6:C"
}, recomposedPositions[1].LinkedVids[0]);
Assert.Equal(new List <string> {
"1:2:G", "1:4:G"
}, recomposedPositions[1].LinkedVids[1]);
}