Entering edit mode
10.8 years ago
Paul
★
1.5k
Dear bio-friends,
Could you please help me how to explain genotyping in my vcf file. I have Illumina data and vcf caller was GATK. My variant frequency (Alt variant freq) is 99.7%. DP = 4622 (AD = 16, 4606) - so I would expect that this sample is alternate homozygous. But when I check PL value, which is - PL = 1655,0,323 - after calculating my likelihood -
REF= G ALT= A
P(D|GG) = 10 ^ -165.5 = small
P(D|AG) = 10 ^ 0 = 1
P(D|AA) = 10 ^ 32.3 = small
we can see it is heterozygous. Can anybody help me how to interpret my result? How it is possible that likelihoods show me heterozygous and coverage and VF show me homozygous?
Here is part of my vcf file:
chr13 32899193 . G A 1625.01 PASS AC=1;AF=0.5;AN=2;DP=4622;QD=0.35;TI=NM_000059;GI=BRCA2;FC=Silent GT:AD:DP:GQ:PL:VF:GQX 0/1:16,4606:5000:99:1655,0,323:0.997:99
Thank you for any explanation.
Paul.
Maybe a quality issue? Someone else will probably have a better answer but given that the Genotyper takes into consideration factors beyond just the number of reads supporting a particular variant, that may be part of the issue. The allelic depth still seems extremely skewed for this to be the answer but that's all I can think of. But there doesn't seem to be any quality filters triggered at that site.
Yeah, true is, that other variants have much more QUAL value... But filter shows PASS and Im not sure how to explain my result.. But thanks for point to QUAL value..
I'd also check for Strand Bias. If a majority of the16 reference allele reads are on one strand, that could impact the genotype call. Checking this region in IGV might give you a good idea.
Thanks for reply.. My filtering is set up for checking Strand Biases.. So there should not be any problem with sb.