public void MergeDbSnp1Kg()
{
//NIR-1262
const string vcfLine =
"1 825069 rs4475692 G A,C . . RS=4475692;RSPOS=825069;dbSNPBuildID=111;SSR=0;SAO=0;VP=0x050100000005170126000100;WGT=1;VC=SNV;SLO;ASP;VLD;G5A;G5;GNO;KGPhase3;CAF=0.3227,.,0.6773;COMMON=1";
const string vcfLine1Kg =
"1 825069 rs4475692 G C 100 PASS AC=3392;AF=0.677316;AN=5008;NS=2504;DP=22495;EAS_AF=0.754;AMR_AF=0.5692;AFR_AF=0.6127;EUR_AF=0.7286;SAS_AF=0.7096;AA=g|||;VT=SNP;EAS_AN=1008;EAS_AC=760;EUR_AN=1006;EUR_AC=733;AFR_AN=1322;AFR_AC=810;AMR_AN=694;AMR_AC=395;SAS_AN=978;SAS_AC=694\tGT";
var dbsnpReader = new DbSnpReader(_renamer);
var sa = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(825069));
foreach (var dbSnpEntry in dbsnpReader.ExtractItem(vcfLine))
{
dbSnpEntry.SetSupplementaryAnnotations(sa);
}
var sa1 = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(825069));
var onekgReader = new OneKGenReader(_renamer);
foreach (var onekgItem in onekgReader.ExtractItems(vcfLine1Kg))
{
onekgItem.SetSupplementaryAnnotations(sa1);
}
sa.MergeSaCreator(sa1);
sa.FinalizePositionalAnnotations();
Assert.Equal("C", sa.SaPosition.GlobalMajorAllele);
Assert.Equal("G", sa.SaPosition.GlobalMinorAllele);
}
public void MergeDbSnp1KpEvsRefMinor()
{
const string vcfLine1 = "1 69428 rs140739101 T G . . RS=140739101;RSPOS=69428;dbSNPBuildID=134;SSR=0;SAO=0;VP=0x050200000a05140026000100;WGT=1;VC=SNV;S3D;NSM;REF;ASP;VLD;KGPhase3;CAF=0.981,0.01897;COMMON=1";
//vcf line is modified
const string vcfLine2 = "1 69428 rs140739101 T G 100 PASS AC=4956;AF=0.989617;AN=5008;NS=2504;DP=17611;EAS_AF=0.003;AMR_AF=0.036;AFR_AF=0.0015;EUR_AF=0.0497;SAS_AF=0.0153;AA=.|||;VT=SNP;EX_TARGET;EAS_AN=1008;EAS_AC=3;EUR_AN=1006;EUR_AC=50;AFR_AN=1322;AFR_AC=2;AMR_AN=694;AMR_AC=25;SAS_AN=978;SAS_AC=15";
const string vcfLine3 = "1 69428 rs140739101 T G . PASS BSNP=dbSNP_134;EA_AC=313,6535;AA_AC=14,3808;TAC=327,10343;MAF=4.5707,0.3663,3.0647;GTS=GG,GT,TT;EA_GTC=92,129,3203;AA_GTC=1,12,1898;GTC=93,141,5101;DP=110;GL=OR4F5;CP=1.0;CG=0.9;AA=T;CA=.;EXOME_CHIP=no;GWAS_PUBMED=.;FG=NM_001005484.1:missense;HGVS_CDNA_VAR=NM_001005484.1:c.338T>G;HGVS_PROTEIN_VAR=NM_001005484.1:p.(F113C);CDS_SIZES=NM_001005484.1:918;GS=205;PH=probably-damaging:0.999;EA_AGE=.;AA_AGE=.";
var sa = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(69428));
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpItem = dbsnpReader.ExtractItem(vcfLine1)[0];
dbSnpItem.SetSupplementaryAnnotations(sa);
var oneKGenItem = _oneKGenReader.ExtractItems(vcfLine2)[0];
oneKGenItem.SetSupplementaryAnnotations(sa);
var evsReader = new EvsReader(_renamer);
var evsItem = evsReader.ExtractItems(vcfLine3)[0];
evsItem.SetSupplementaryAnnotations(sa);
sa.FinalizePositionalAnnotations();
var dbSnp = sa.SaPosition.AlleleSpecificAnnotations["G"].Annotations[DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)] as DbSnpAnnotation;
Assert.NotNull(dbSnp);
Assert.Equal(new List <long> {
140739101
}, dbSnp.DbSnp);
Assert.Equal(true, sa.SaPosition.IsRefMinorAllele);
}
public void MergeSnvAndDeletion()
{
// NIR-906
const string vcfLine1 =
"1 862389 rs6693546 A G . . RS=6693546;RSPOS=862389;dbSNPBuildID=116;SSR=0;SAO=0;VP=0x05010008000515013e000100;WGT=1;VC=SNV;SLO;INT;ASP;VLD;G5;GNO;KGPhase1;KGPhase3;CAF=0.3171,0.6829;COMMON=1";
const string vcfLine2 = "1 862388 rs534606253 GA G . . RS=534606253;RSPOS=862389;dbSNPBuildID=142;SSR=0;SAO=0;VP=0x050000080005040024000200;WGT=1;VC=DIV;INT;ASP;VLD;KGPhase3;CAF=0.996,0.003994;COMMON=1";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntry1 = dbsnpReader.ExtractItem(vcfLine1)[0];
var dbSnpEntry2 = dbsnpReader.ExtractItem(vcfLine2)[0];
var sa = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(862389));
dbSnpEntry1.SetSupplementaryAnnotations(sa);
var sa2 = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(862389));
var additionalEntry = dbSnpEntry2.SetSupplementaryAnnotations(sa2);
additionalEntry.SetSupplementaryAnnotations(sa2);
sa.MergeSaCreator(sa2);
sa.FinalizePositionalAnnotations();
Assert.Equal("G", sa.SaPosition.GlobalMajorAllele);
Assert.Equal("0.6829", sa.SaPosition.GlobalMajorAlleleFrequency);
Assert.Equal("A", sa.SaPosition.GlobalMinorAllele);
Assert.Equal("0.3171", sa.SaPosition.GlobalMinorAlleleFrequency);
}
public void DeletionAndSnvMerge()
{
// NIR-906
const string vcfLine1 =
"2 193187631 rs774176075 TGTTG T . . RS=774176075;RSPOS=193187632;dbSNPBuildID=144;SSR=0;SAO=0;VP=0x050000000005000002000200;WGT=1;VC=DIV;ASP";
const string vcfLine2 = "2 193187632 rs2592266 G T . . RS=2592266;RSPOS=193187632;dbSNPBuildID=100;SSR=0;SAO=0;VP=0x050000000005150026000100;WGT=1;VC=SNV;ASP;VLD;G5;KGPhase3;CAF=0.01937,0.9806;COMMON=1";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntry1 = dbsnpReader.ExtractItem(vcfLine1)[0];
var dbSnpEntry2 = dbsnpReader.ExtractItem(vcfLine2)[0];
var sa = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(193187632));
var additionalEntry = dbSnpEntry1.SetSupplementaryAnnotations(sa);
additionalEntry.SetSupplementaryAnnotations(sa);
var sa2 = new SupplementaryPositionCreator(new SupplementaryAnnotationPosition(193187632));
dbSnpEntry2.SetSupplementaryAnnotations(sa2);
sa.MergeSaCreator(sa2);
sa.FinalizePositionalAnnotations();
Assert.Equal("T", sa.SaPosition.GlobalMajorAllele);
Assert.Equal("0.9806", sa.SaPosition.GlobalMajorAlleleFrequency);
Assert.Equal("G", sa.SaPosition.GlobalMinorAllele);
Assert.Equal("0.01937", sa.SaPosition.GlobalMinorAlleleFrequency);
}
public void RwDbsnpGlobalAlleles()
{
//NIR-1262
var randomPath = Path.Combine(Path.GetTempPath(), Path.GetRandomFileName());
// create our expected data source versions
var dbSnpVersion = new DataSourceVersion("dbSNP", "147", DateTime.Parse("2016-07-26").Ticks);
var expectedDataSourceVersions = new List <DataSourceVersion> {
dbSnpVersion
};
const string vcfLine1 =
"2 141724543 rs112783784 A C,T . . RS=112783784;RSPOS=141724543;dbSNPBuildID=132;SSR=0;SAO=0;VP=0x050100080015140136000100;WGT=1;VC=SNV;SLO;INT;OTH;ASP;VLD;GNO;KGPhase1;KGPhase3;CAF=0.04113,0.9105,0.04832;COMMON=1";
const string vcfLine2 =
"2 141724543 rs4300776 A C . . RS=4300776;RSPOS=141724543;dbSNPBuildID=111;SSR=0;SAO=0;VP=0x050100080015000102000100;WGT=1;VC=SNV;SLO;INT;OTH;ASP;GNO;CAF=0.04113,0.9105;COMMON=1";
var sa = new SupplementaryAnnotationPosition(141724543);
var saCreator = new SupplementaryPositionCreator(sa);
var dbsnpReader = new DbSnpReader(_renamer);
foreach (var dbSnpItem in dbsnpReader.ExtractItem(vcfLine1))
{
dbSnpItem.SetSupplementaryAnnotations(saCreator);
}
foreach (var dbSnpItem in dbsnpReader.ExtractItem(vcfLine2))
{
dbSnpItem.SetSupplementaryAnnotations(saCreator);
}
// write the supplementary annotation file
using (var writer = new SupplementaryAnnotationWriter(randomPath, "chr1", expectedDataSourceVersions))
{
saCreator.FinalizePositionalAnnotations();
writer.Write(saCreator, sa.ReferencePosition);
}
using (var reader = new SupplementaryAnnotationReader(randomPath))
{
// extract the three annotations
var observedAnnotation = reader.GetAnnotation(141724543) as SupplementaryAnnotationPosition;
Assert.NotNull(observedAnnotation);
Assert.Equal("C", observedAnnotation.GlobalMajorAllele);
Assert.Equal("0.9105", observedAnnotation.GlobalMajorAlleleFrequency);
Assert.Equal("T", observedAnnotation.GlobalMinorAllele);
Assert.Equal("0.04832", observedAnnotation.GlobalMinorAlleleFrequency);
}
File.Delete(randomPath);
File.Delete(randomPath + ".idx");
}
public void RefAlleleMinor()
{
var sa = new SupplementaryPositionCreator();
foreach (var oneKitem in _oneKGenReader.ExtractItems(VcfLine2))
{
oneKitem.SetSupplementaryAnnotations(sa);
}
sa.FinalizePositionalAnnotations();
Assert.Equal(sa.IsRefMinor(), true);
}
public void MissingRefMinor()
{
const string vcfLine =
"1 15274 rs62636497 A G,T 100 PASS AC=1739,3210;AF=0.347244,0.640974;AN=5008;NS=2504;DP=23255;EAS_AF=0.4812,0.5188;AMR_AF=0.2752,0.7205;AFR_AF=0.323,0.6369;EUR_AF=0.2922,0.7078;SAS_AF=0.3497,0.6472;AA=g|||;VT=SNP;MULTI_ALLELIC";
var sa = new SupplementaryPositionCreator();
foreach (var oneKitems in _oneKGenReader.ExtractItems(vcfLine))
{
oneKitems.SetSupplementaryAnnotations(sa);
}
sa.FinalizePositionalAnnotations();
Assert.True(sa.IsRefMinor());
}
public void RefGlobalMajor()
{
// NIR-942
const string vcfLine =
"1 1242707 rs2274262 A G,T . . RS=2274262;RSPOS=1242707;RV;dbSNPBuildID=100;SSR=0;SAO=0;VP=0x0501004a000507013e000100;WGT=1;VC=SNV;SLO;U5;INT;R5;ASP;VLD;G5A;G5;GNO;KGPhase1;KGPhase3;CAF=0.4,0.4,0.2;COMMON=1";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntry = dbsnpReader.ExtractItem(vcfLine)[0];
var sa = new SupplementaryAnnotationPosition(1242707);
var saCreator = new SupplementaryPositionCreator(sa);
dbSnpEntry.SetSupplementaryAnnotations(saCreator);
saCreator.FinalizePositionalAnnotations();
Assert.Equal("A", sa.GlobalMajorAllele);
Assert.Equal("G", sa.GlobalMinorAllele);
}
public void DisregardZeroFreq()
{
const string vcfLine =
"1 241369 rs11490246 C T . . RS=11490246;RSPOS=241369;dbSNPBuildID=120;SSR=0;SAO=0;VP=0x050100000005000126000100;WGT=1;VC=SNV;SLO;ASP;GNO;KGPhase3;CAF=0,1;COMMON=0";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntry = dbsnpReader.ExtractItem(vcfLine)[0];
var sa = new SupplementaryAnnotationPosition(828);
var saCreator = new SupplementaryPositionCreator(sa);
dbSnpEntry.SetSupplementaryAnnotations(saCreator);
saCreator.FinalizePositionalAnnotations();
Assert.Equal("T", sa.GlobalMajorAllele);
Assert.Equal("1", sa.GlobalMajorAlleleFrequency);
Assert.Null(sa.GlobalMinorAllele);
Assert.Null(sa.GlobalMinorAlleleFrequency);
}
public void MultiEntryMixedVariant()
{
var sa1 = new SupplementaryPositionCreator();
var oneKItem1 = _oneKGenReader.ExtractItems(VcfLine5)[0];
var oneKItem2 = _oneKGenReader.ExtractItems(VcfLine6)[0];
oneKItem1.SetSupplementaryAnnotations(sa1);
// additional items are ignored since they cannot be SNVs
var sa2 = new SupplementaryPositionCreator();
oneKItem2.SetSupplementaryAnnotations(sa2);
sa1.MergeSaCreator(sa2);
sa1.FinalizePositionalAnnotations();
Assert.True(sa1.IsRefMinor());
}
public void MissingDbsnpId()
{
// refactorSA. Annotation for C is missing in the database. have to debug that.
const string vcfLine =
"X 21505833 rs12395602 G A,C,T . . RS=12395602;RSPOS=21505833;dbSNPBuildID=120;SSR=0;SAO=0;VP=0x05010008000505051f000101;WGT=1;VC=SNV;SLO;INT;ASP;VLD;G5;HD;GNO;KGPhase1";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntries = dbsnpReader.ExtractItem(vcfLine);
var sa = new SupplementaryAnnotationPosition(21505833);
var saCreator = new SupplementaryPositionCreator(sa);
foreach (var dbSnpEntry in dbSnpEntries)
{
dbSnpEntry.SetSupplementaryAnnotations(saCreator);
}
saCreator.FinalizePositionalAnnotations();
var dbSnpA =
sa.AlleleSpecificAnnotations["A"].Annotations[DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)] as
DbSnpAnnotation;
Assert.NotNull(dbSnpA);
var dbSnpC =
sa.AlleleSpecificAnnotations["C"].Annotations[DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)] as
DbSnpAnnotation;
Assert.NotNull(dbSnpC);
var dbSnpT =
sa.AlleleSpecificAnnotations["T"].Annotations[DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)] as
DbSnpAnnotation;
Assert.NotNull(dbSnpT);
Assert.Equal(12395602, dbSnpA.DbSnp[0]);
Assert.Equal(12395602, dbSnpC.DbSnp[0]);
Assert.Equal(12395602, dbSnpT.DbSnp[0]);
}
public void NoMinorAllele1()
{
const string vcfLine =
"17 828 rs62053745 T C . . RS=62053745;RSPOS=828;dbSNPBuildID=129;SSR=0;SAO=0;VP=0x050100080005140136000100;WGT=1;VC=SNV;SLO;INT;ASP;VLD;GNO;KGPhase1;KGPhase3;CAF=0.7424,.;COMMON=1";
var dbsnpReader = new DbSnpReader(_renamer);
var dbSnpEntry = dbsnpReader.ExtractItem(vcfLine)[0];
var sa = new SupplementaryAnnotationPosition(828);
var saCreator = new SupplementaryPositionCreator(sa);
dbSnpEntry.SetSupplementaryAnnotations(saCreator);
saCreator.FinalizePositionalAnnotations();
Assert.Equal("T", sa.GlobalMajorAllele);
Assert.Equal("0.7424", sa.GlobalMajorAlleleFrequency);
Assert.Null(sa.GlobalMinorAllele);
Assert.Null(sa.GlobalMinorAlleleFrequency);
}
public void SpuriousRefMinor()
{
// NIR-903
const string vcfLine =
"2 190634102 rs531674661;rs1225108 A AC,C 100 PASS AC=18,4905;AF=0.00359425,0.979433;AN=5008;NS=2504;DP=14024;EAS_AF=0.001,0.997;AMR_AF=0.0043,0.9899;AFR_AF=0,0.9402;EUR_AF=0.004,0.996;SAS_AF=0.0102,0.9898;VT=SNP,INDEL;MULTI_ALLELIC";
var sa = new SupplementaryPositionCreator();
var sa1 = new SupplementaryPositionCreator();
var oneKitems = _oneKGenReader.ExtractItems(vcfLine);
oneKitems[0].SetSupplementaryAnnotations(sa);
oneKitems[1].SetSupplementaryAnnotations(sa1);
sa.FinalizePositionalAnnotations();
sa1.FinalizePositionalAnnotations();
Assert.False(sa.IsRefMinor());
Assert.True(sa1.IsRefMinor());
}
/// <summary>
/// writes the annotations to the current database file
/// </summary>
public void Write(SupplementaryPositionCreator spCreator, int referencePos, bool finalizePositinalAnnotation = true)
{
if (finalizePositinalAnnotation)
{
spCreator.FinalizePositionalAnnotations();
}
if (spCreator.IsEmpty())
{
return;
}
// add this entry to the index
var currentOffset = _stream.Position;
_index.Add((uint)referencePos, (uint)currentOffset, spCreator.IsRefMinor());
if (spCreator.IsRefMinor())
{
RefMinorCount++;
}
spCreator.WriteAnnotation(_writer);
}