the Allele Count (AC) tag is incorrect

naarkhoo

I am facing this error, when I try to validate the variants (ValidateVariants) of a vcf file which is produced through GATK just after UnifiedGenotyper. I am using GenomeAnalysisTK-2.3-6-gebbba25 and dbsnp_137.hg19.vcf. These variants are annotated by DepthOfCoverage, aplotypeScore, ,InbreedingCoeff and LowMQ ...

Basically, I generate two VCF files using UnifiedGenotyper separately, one for SNP and the other for INDEL.

the error for both is about the Allele Count (AC) tag: ##### ERROR MESSAGE: File F93.snp.vcf fails strict validation: the Allele Count (AC) tag is incorrect for the record at position chr1:1225579, 1 vs. 1

I appreciate your comments,

Geraldine_VdAuwera

Can you upgrade to the latest version and see if the error persists?

naarkhoo

I tried GenomeAnalysisTK-2.3-9-ge5ebf34, the error still persists ! the Allele Count (AC) tag is incorrect for the record at position chrM:302, 2 vs. 2

Geraldine_VdAuwera

Ok. Can you post the VCF record where the error occurs (chrM:302)?

naarkhoo

chrM    302 rs66492218  AC  ACC,A   2239.04 PASS    AC=3,2;AF=0.375,0.250;AN=8;BaseQRankSum=0.822;DB;DP=195;FS=3.608;HaplotypeScore=45.0534;IndelType=MULTIALLELIC_INDEL;LowMQ=0.0000,0.0000,195;MLEAC=3,2;MLEAF=0.375,0.250;MQ=58.74;MQ0=0;MQRankSum=1.342;QD=11.48;RPA=8,9,7;RU=C;ReadPosRankSum=-1.658;STR;set=variant   

Geraldine_VdAuwera

Can you please post the complete record? This is missing the format field and sample values.

naarkhoo

chrM    302 rs66492218  AC  ACC,A   2239.04 PASS    AC=3,2;AF=0.375,0.250;AN=8;BaseQRankSum=0.822;DB;DP=195;FS=3.608;HaplotypeScore=45.0534;IndelType=MULTIALLELIC_INDEL;LowMQ=0.0000,0.0000,195;MLEAC=3,2;MLEAF=0.375,0.250;MQ=58.74;MQ0=0;MQRankSum=1.342;QD=11.48;RPA=8,9,7;RU=C;ReadPosRankSum=-1.658;STR;set=variant   GT:AD:DP:GQ:PL  ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. ./. 0/1:41,10,2:66:99:186,0,1210,211,1129,1608  0/1:3,34,0:47:21:987,0,21,1084,127,1467 2/2:0,0,36:52:99:1179,1264,1506,108,108,0   0/1:20,1,0:30:2:2,0,542,70,580,739

Geraldine_VdAuwera

Ah, there you go -- the counts are indeed incorrect. The record includes 3 samples with the first allele but only 1 of the second, unlike what is reported in the AC tag. Did you exclude some samples after processing, perhaps?

naarkhoo

no ! I didn't ! I used UnifiedGenotyper separably for calling SNP and indels from 93 samples ; then I did filtration

for indel

--filterExpression "QD < 2.0" \
--filterName "QDFilter" \
--filterExpression "ReadPosRankSum < -20.0" \
--filterName "ReadPosFilter" \
--filterExpression "FS > 200.0" \
--filterName "FSFilter" \
--filterExpression "MQ0 >= 4 && ((MQ0 / (1.0 * DP)) > 0.1)" \
--filterName "HARD_TO_VALIDATE" \
--filterExpression "QUAL < 30.0 || DP < 6 || DP > 5000 || HRun > 5" \
--filterName "QualFilter"

for SNP:

--clusterSize 3 \
--clusterWindowSize 10 \
--filterExpression "QD < 2.0" \
--filterName "QDFilter" \
--filterExpression "MQ < 40.0" \
--filterName "MQFilter" \
--filterExpression "FS > 60.0" \
--filterName "FSFilter" \
--filterExpression "HaplotypeScore > 13.0" \
--filterName "HaplotypeScoreFilter" \
--filterExpression "MQRankSum < -12.5" \
--filterName "MQRankSumFilter" \
--filterExpression "ReadPosRankSum < -8.0" \
--filterName "ReadPosRankSumFilter" \
--filterExpression "QUAL < 30.0 || DP < 6 || DP > 5000 || HRun > 5" \

and then I combined these two, using CombineVariants; as you see, I didn't remove any sample.

Geraldine_VdAuwera

I see -- Just to be clear, are you getting the error when you run ValidateVariants on the files that come straight out of the UnifiedGenotyper, or on the combined VCF that results from CombineVariants?

Geraldine_VdAuwera

Ah, actually I was mistaken -- the sample that is 2/2 counts for two alleles, so the tag is in fact correct. We'll have a closer look at this.

naarkhoo

Surprisingly, I don't see the error again ! sounds like miracle :D , now I am getting something like File F93.all.vcf fails strict validation: the rsID rs35614524 for the record at position chr9:139565479 is not in dbSNP ... I haven't start to pray for it to get fix automatically

Geraldine_VdAuwera

Hmm, that's odd. Well, the new version will be out very soon, which is more thoroughly tested -- hopefully you won't suffer these weird phantom bugs then.