public void IsFivePrimeUtrVariant(bool onReverseStrand, bool withinCdna, bool beforeCoding, bool afterCoding,
bool expectedResult)
{
var positionalEffect = new TranscriptPositionalEffect
{
BeforeCoding = beforeCoding,
AfterCoding = afterCoding,
WithinCdna = withinCdna
};
var variant = new Mock <ISimpleVariant>();
var transcript = new Mock <ITranscript>();
variant.SetupGet(x => x.AltAllele).Returns("G");
variant.SetupGet(x => x.RefAllele).Returns("C");
var variantEffect = new VariantEffect(positionalEffect, variant.Object, transcript.Object, "", "", "", "",
null, "", "");
var gene = new Mock <IGene>();
transcript.SetupGet(x => x.Gene).Returns(gene.Object);
gene.SetupGet(x => x.OnReverseStrand).Returns(onReverseStrand);
var translation = new Mock <ITranslation>();
transcript.SetupGet(x => x.Translation).Returns(translation.Object);
Assert.Equal(expectedResult, variantEffect.IsFivePrimeUtrVariant());
}
private static List <ConsequenceTag> GetConsequences(ITranscript transcript, IVariant variant,
VariantEffect variantEffect)
{
var featureEffect = new FeatureVariantEffects(transcript, variant.Type, variant,
variant.Behavior.StructuralVariantConsequence);
var consequence = new Consequences(variantEffect, featureEffect);
consequence.DetermineSmallVariantEffects();
return(consequence.GetConsequences());
}
private static VariantEffect GetVariantEffect(ITranscript transcript, ISimpleVariant variant, IMappedPositions mappedPositions, string refAminoAcids, string altAminoAcids, string refCodons, string altCodons, bool insertionInStartAndNoImpact)
{
var positionalEffect = new TranscriptPositionalEffect();
positionalEffect.DetermineIntronicEffect(transcript.Introns, variant, variant.Type);
positionalEffect.DetermineExonicEffect(transcript, variant, mappedPositions, variant.AltAllele, insertionInStartAndNoImpact);
var variantEffect = new VariantEffect(positionalEffect, variant, transcript, refAminoAcids,
altAminoAcids,
refCodons, altCodons, mappedPositions.ProteinInterval.Start);
return(variantEffect);
}
/// <summary>
/// constructor
/// </summary>
public HgvsProteinNomenclature(VariantEffect variantEffect, TranscriptAnnotation ta, Transcript transcript,
VariantFeature variant, ICompressedSequence compressedSequence, AminoAcids aminoAcids)
{
_variantEffect = variantEffect;
_ta = ta;
_transcript = transcript;
_variant = variant;
_compressedSequence = compressedSequence;
_aminoAcids = aminoAcids;
_hgvsNotation = new HgvsNotation(_ta.ReferenceAminoAcids, _ta.AlternateAminoAcids,
FormatUtilities.CombineIdAndVersion(_transcript.Translation.ProteinId, _transcript.Translation.ProteinVersion),
_ta.ProteinBegin, _ta.ProteinEnd);
}
AltAminoAcids, string RefCodons, string AltCodons, string TranscriptAltAllele, string TranscriptRefAllele) AnnotateTranscript(ITranscript transcript, ISimpleVariant variant, AminoAcids aminoAcids, ISequence refSequence)
{
bool onReverseStrand = transcript.Gene.OnReverseStrand;
var start = MappedPositionUtilities.FindRegion(transcript.TranscriptRegions, variant.Start);
var end = MappedPositionUtilities.FindRegion(transcript.TranscriptRegions, variant.End);
var position = GetMappedPosition(transcript.TranscriptRegions, start.Region, start.Index, end.Region,
end.Index, variant, onReverseStrand, transcript.Translation?.CodingRegion, transcript.StartExonPhase,
variant.Type == VariantType.insertion);
var codingSequence = GetCodingSequence(transcript, refSequence);
var cdnaSequence = GetCdnaSequence(transcript, refSequence);
string transcriptAltAllele = HgvsUtilities.GetTranscriptAllele(variant.AltAllele, onReverseStrand);
var codons = Codons.GetCodons(transcriptAltAllele, position.CdsStart, position.CdsEnd, position.ProteinStart, position.ProteinEnd, codingSequence);
var aa = aminoAcids.Translate(codons.Reference, codons.Alternate);
(aa, position.ProteinStart, position.ProteinEnd) = TryTrimAminoAcidsAndUpdateProteinPositions(aa, position.ProteinStart, position.ProteinEnd);
(position.CoveredCdnaStart, position.CoveredCdnaEnd) = transcript.TranscriptRegions.GetCoveredCdnaPositions(position.CdnaStart, start.Index, position.CdnaEnd, end.Index, onReverseStrand);
(position.CoveredCdsStart, position.CoveredCdsEnd, position.CoveredProteinStart, position.CoveredProteinEnd) = MappedPositionUtilities.GetCoveredCdsAndProteinPositions(position.CoveredCdnaStart, position.CoveredCdnaEnd, transcript.StartExonPhase, transcript.Translation?.CodingRegion);
var transcriptRefAllele = GetTranscriptRefAllele(position, cdnaSequence, variant, onReverseStrand);
SequenceChange coveredAa;
// only generate the covered version of ref & alt alleles when CDS start/end is -1
if (position.CdsStart == -1 || position.CdsEnd == -1)
{
coveredAa = GetCoveredAa(aminoAcids, transcriptAltAllele, position.CoveredCdsStart, position.CoveredCdsEnd, position.CoveredProteinStart, position.CoveredProteinEnd, codingSequence);
(coveredAa, position.CoveredProteinStart, position.CoveredProteinEnd) = TryTrimAminoAcidsAndUpdateProteinPositions(coveredAa, position.CoveredProteinStart, position.CoveredProteinEnd);
}
else
{
coveredAa = aa;
position.CoveredProteinStart = position.ProteinStart;
position.CoveredProteinEnd = position.ProteinEnd;
}
var positionalEffect = GetPositionalEffect(transcript, variant, position, aa.Reference, aa.Alternate,
position.CoveredCdnaStart, position.CoveredCdnaEnd, position.CoveredCdsStart, position.CoveredCdsEnd);
var variantEffect = new VariantEffect(positionalEffect, variant, transcript, aa.Reference, aa.Alternate,
codons.Reference, codons.Alternate, position.ProteinStart, coveredAa.Reference, coveredAa.Alternate);
return(variantEffect, position, aa.Reference, aa.Alternate, codons.Reference, codons.Alternate, transcriptAltAllele, transcriptRefAllele);
}
public void IsStopLost_DeletionOverStopCodon_ReturnTrue()
{
var positionalEffect = new TranscriptPositionalEffect
{
BeforeCoding = false,
AfterCoding = true,
WithinCdna = true
};
var variant = new Mock <ISimpleVariant>();
variant.SetupGet(x => x.AltAllele).Returns("ATAGCCC");
variant.SetupGet(x => x.RefAllele).Returns("A");
var variantEffect = new VariantEffect(positionalEffect, variant.Object, null, "", "", "", "",
null, "*", "X");
Assert.True(variantEffect.IsStopLost());
}
public void IsStartRetainedVariant(int proteinBegin, string refAminoAcids, string altAminoAcids, string refAllele, string altAllele, bool isStartRetained)
{
var variant = new Mock <ISimpleVariant>();
var transcript = new Mock <ITranscript>();
variant.SetupGet(x => x.AltAllele).Returns(refAllele);
variant.SetupGet(x => x.RefAllele).Returns(altAllele);
var variantEffect = new VariantEffect(null, variant.Object, transcript.Object, refAminoAcids, altAminoAcids, "", "",
proteinBegin, refAminoAcids, altAminoAcids);
if (isStartRetained)
{
Assert.True(variantEffect.IsStartRetained());
}
else
{
Assert.False(variantEffect.IsStartRetained());
}
}
AltAminoAcids, string RefCodons, string AltCodons, string TranscriptAltAllele) AnnotateTranscript(ITranscript transcript, ISimpleVariant variant, AminoAcids aminoAcids,
ISequence refSequence)
{
var onReverseStrand = transcript.Gene.OnReverseStrand;
var start = MappedPositionUtilities.FindRegion(transcript.TranscriptRegions, variant.Start);
var end = MappedPositionUtilities.FindRegion(transcript.TranscriptRegions, variant.End);
var position = GetMappedPosition(transcript.TranscriptRegions, start.Region, start.Index, end.Region,
end.Index, variant, onReverseStrand, transcript.Translation?.CodingRegion, transcript.StartExonPhase,
variant.Type == VariantType.insertion);
var transcriptRefAllele = HgvsUtilities.GetTranscriptAllele(variant.RefAllele, onReverseStrand);
var transcriptAltAllele = HgvsUtilities.GetTranscriptAllele(variant.AltAllele, onReverseStrand);
var codingSequence = transcript.Translation == null
? null
: new CodingSequence(refSequence, transcript.Translation.CodingRegion, transcript.TranscriptRegions,
transcript.Gene.OnReverseStrand, transcript.StartExonPhase);
var codons = Codons.GetCodons(transcriptRefAllele, transcriptAltAllele, position.CdsStart, position.CdsEnd,
position.ProteinStart, position.ProteinEnd, codingSequence);
var coveredCdna = transcript.TranscriptRegions.GetCoveredCdnaPositions(position.CdnaStart, start.Index,
position.CdnaEnd, end.Index, onReverseStrand);
var coveredCds = MappedPositionUtilities.GetCoveredCdsPositions(coveredCdna.Start, coveredCdna.End,
transcript.StartExonPhase, transcript.Translation?.CodingRegion);
var aa = aminoAcids.Translate(codons.Reference, codons.Alternate);
var positionalEffect = GetPositionalEffect(transcript, variant, position, aa.Reference, aa.Alternate,
coveredCdna.Start, coveredCdna.End, coveredCds.Start, coveredCds.End);
var variantEffect = new VariantEffect(positionalEffect, variant, transcript, aa.Reference, aa.Alternate,
codons.Reference, codons.Alternate, position.ProteinStart);
return(variantEffect, position, aa.Reference, aa.Alternate, codons.Reference, codons.Alternate,
transcriptAltAllele);
}
public void IsCodingSequenceVariant_WithMoreSpecificConsequence_ReturnFalse(ConsequenceTag ct)
{
var positionalEffect = new TranscriptPositionalEffect
{
BeforeCoding = false,
AfterCoding = true,
WithinCdna = true
};
var cache = new VariantEffectCache();
cache.Add(ct, true);
var variant = new Mock <ISimpleVariant>();
variant.SetupGet(x => x.AltAllele).Returns("ATAGCCC");
variant.SetupGet(x => x.RefAllele).Returns("A");
var variantEffect = new VariantEffect(positionalEffect, variant.Object, null, "", "", "", "",
null, null, null, cache);
Assert.False(variantEffect.IsCodingSequenceVariant());
}
public void IsSpliceDonorVariant(bool onReverseStrand, bool isStartSpliceSite, bool isEndSpliceSite, bool expectedResult)
{
var positionalEffect = new TranscriptPositionalEffect
{
IsStartSpliceSite = isStartSpliceSite,
IsEndSpliceSite = isEndSpliceSite
};
var variant = new Mock <ISimpleVariant>();
var transcript = new Mock <ITranscript>();
variant.SetupGet(x => x.AltAllele).Returns("G");
variant.SetupGet(x => x.RefAllele).Returns("C");
var variantEffect = new VariantEffect(positionalEffect, variant.Object, transcript.Object, "", "", "", "",
null, "", "");
var gene = new Mock <IGene>();
transcript.SetupGet(x => x.Gene).Returns(gene.Object);
gene.SetupGet(x => x.OnReverseStrand).Returns(onReverseStrand);
Assert.Equal(expectedResult, variantEffect.IsSpliceDonorVariant());
}
public static string GetHgvsProteinAnnotation(
ITranscript transcript,
string refAminoAcids,
string altAminoAcids,
string transcriptAltAllele,
IMappedPosition position,
VariantEffect variantEffect,
ISimpleVariant variant,
ISequence refSequence,
string hgvscNotation,
bool isMitochondrial)
{
if (IsHgvspNull(transcriptAltAllele, position.CdsStart, position.CdsEnd, variant, hgvscNotation))
{
return(null);
}
var peptideSeq = transcript.Translation.PeptideSeq;
// Amino acid seq should never go past the stop codon
refAminoAcids = !refAminoAcids.EndsWith(AminoAcids.StopCodon) && refAminoAcids.Contains(AminoAcids.StopCodon)
? refAminoAcids.OptimizedSplit(AminoAcids.StopCodon[0])[0] + AminoAcids.StopCodon
: refAminoAcids;
int proteinStart = position.ProteinStart;
HgvsUtilities.ShiftAndRotateAlleles(ref proteinStart, ref refAminoAcids, ref altAminoAcids, peptideSeq);
var end = proteinStart + refAminoAcids.Length - 1;
var refAbbreviation = AminoAcids.GetAbbreviations(refAminoAcids);
var altAbbreviation = AminoAcids.GetAbbreviations(altAminoAcids);
var proteinId = transcript.Translation.ProteinId.WithVersion;
var proteinChange = GetProteinChange(proteinStart, refAminoAcids, altAminoAcids, peptideSeq, variantEffect);
// ReSharper disable once SwitchStatementMissingSomeCases
switch (proteinChange)
{
case ProteinChange.Substitution:
return(HgvspNotation.GetSubstitutionNotation(proteinId, proteinStart, refAbbreviation, altAbbreviation));
case ProteinChange.Unknown:
return(HgvspNotation.GetUnknownNotation(proteinId, proteinStart, end, refAbbreviation, altAbbreviation));
case ProteinChange.Deletion:
return(HgvspNotation.GetDeletionNotation(proteinId, proteinStart, end, refAbbreviation, variantEffect.IsStopGained()));
case ProteinChange.Duplication:
proteinStart -= altAminoAcids.Length;
return(HgvspNotation.GetDuplicationNotation(proteinId, proteinStart, end, altAbbreviation));
case ProteinChange.Frameshift:
return(GetHgvsFrameshiftNotation(refSequence, position.CdsStart, position.CdsEnd, transcriptAltAllele,
transcript, isMitochondrial, proteinId, proteinStart, end));
case ProteinChange.None:
return(HgvspNotation.GetSilentNotation(hgvscNotation, proteinStart, refAbbreviation, variantEffect.IsStopRetained()));
case ProteinChange.DelIns:
return(HgvspNotation.GetDelInsNotation(proteinId, proteinStart, end, refAbbreviation, altAbbreviation));
case ProteinChange.Insertion:
Swap.Int(ref proteinStart, ref end);
return(HgvspNotation.GetInsertionNotation(proteinId, proteinStart, end, altAbbreviation, peptideSeq));
case ProteinChange.Extension:
var altPeptideSequence = HgvsUtilities.GetAltPeptideSequence(refSequence, position.CdsStart, position.CdsEnd,
transcriptAltAllele, transcript, isMitochondrial);
altAbbreviation = proteinStart <= altPeptideSequence.Length ? AminoAcids.ConvertAminoAcidToAbbreviation(altPeptideSequence[proteinStart - 1]): "Ter";
var countToStop = HgvsUtilities.GetNumAminoAcidsUntilStopCodon(altPeptideSequence, peptideSeq, proteinStart - 1, false);
return(HgvspNotation.GetExtensionNotation(proteinId, proteinStart, refAbbreviation, altAbbreviation, countToStop));
case ProteinChange.StartLost:
return(HgvspNotation.GetStartLostNotation(proteinId, proteinStart, end, refAbbreviation));
}
return(null);
}
public void DetermineSmallVariantEffects_tier3()
{
var cache = new VariantEffectCache();
cache.Add(ConsequenceTag.mature_miRNA_variant, false);
cache.Add(ConsequenceTag.splice_donor_variant, true);
cache.Add(ConsequenceTag.splice_acceptor_variant, true);
cache.Add(ConsequenceTag.stop_gained, true);
cache.Add(ConsequenceTag.frameshift_variant, true);
cache.Add(ConsequenceTag.stop_lost, true);
cache.Add(ConsequenceTag.start_lost, true);
cache.Add(ConsequenceTag.inframe_insertion, true);
cache.Add(ConsequenceTag.inframe_deletion, true);
cache.Add(ConsequenceTag.missense_variant, true);
cache.Add(ConsequenceTag.protein_altering_variant, true);
cache.Add(ConsequenceTag.splice_region_variant, true);
cache.Add(ConsequenceTag.incomplete_terminal_codon_variant, true);
cache.Add(ConsequenceTag.stop_retained_variant, true);
cache.Add(ConsequenceTag.synonymous_variant, true);
cache.Add(ConsequenceTag.coding_sequence_variant, true);
cache.Add(ConsequenceTag.five_prime_UTR_variant, true);
cache.Add(ConsequenceTag.three_prime_UTR_variant, true);
cache.Add(ConsequenceTag.non_coding_transcript_exon_variant, true);
cache.Add(ConsequenceTag.intron_variant, true);
cache.Add(ConsequenceTag.NMD_transcript_variant, true);
cache.Add(ConsequenceTag.non_coding_transcript_variant, true);
var simpleVariant = new Mock <ISimpleVariant>();
simpleVariant.SetupGet(x => x.RefAllele).Returns("G");
simpleVariant.SetupGet(x => x.AltAllele).Returns("C");
var positionalEffect = new TranscriptPositionalEffect
{
IsWithinIntron = true
};
var variantEffect = new VariantEffect(positionalEffect, simpleVariant.Object, null, null, null, null, null, null, cache);
var featureEffect = new Mock <IFeatureVariantEffects>();
featureEffect.Setup(x => x.Ablation()).Returns(false);
featureEffect.Setup(x => x.Amplification()).Returns(false);
featureEffect.Setup(x => x.Truncation()).Returns(true);
featureEffect.Setup(x => x.Elongation()).Returns(true);
var consequence = new Consequences(variantEffect, featureEffect.Object);
consequence.DetermineSmallVariantEffects();
var observedConsequence = consequence.GetConsequences();
Assert.Equal(ConsequenceTag.splice_donor_variant, observedConsequence[0]);
Assert.Equal(ConsequenceTag.splice_acceptor_variant, observedConsequence[1]);
Assert.Equal(ConsequenceTag.stop_gained, observedConsequence[2]);
Assert.Equal(ConsequenceTag.frameshift_variant, observedConsequence[3]);
Assert.Equal(ConsequenceTag.stop_lost, observedConsequence[4]);
Assert.Equal(ConsequenceTag.start_lost, observedConsequence[5]);
Assert.Equal(ConsequenceTag.inframe_insertion, observedConsequence[6]);
Assert.Equal(ConsequenceTag.inframe_deletion, observedConsequence[7]);
Assert.Equal(ConsequenceTag.missense_variant, observedConsequence[8]);
Assert.Equal(ConsequenceTag.protein_altering_variant, observedConsequence[9]);
Assert.Equal(ConsequenceTag.splice_region_variant, observedConsequence[10]);
Assert.Equal(ConsequenceTag.incomplete_terminal_codon_variant, observedConsequence[11]);
Assert.Equal(ConsequenceTag.stop_retained_variant, observedConsequence[12]);
Assert.Equal(ConsequenceTag.synonymous_variant, observedConsequence[13]);
Assert.Equal(ConsequenceTag.coding_sequence_variant, observedConsequence[14]);
Assert.Equal(ConsequenceTag.five_prime_UTR_variant, observedConsequence[15]);
Assert.Equal(ConsequenceTag.three_prime_UTR_variant, observedConsequence[16]);
Assert.Equal(ConsequenceTag.non_coding_transcript_exon_variant, observedConsequence[17]);
Assert.Equal(ConsequenceTag.intron_variant, observedConsequence[18]);
Assert.Equal(ConsequenceTag.NMD_transcript_variant, observedConsequence[19]);
Assert.Equal(ConsequenceTag.non_coding_transcript_variant, observedConsequence[20]);
}