Variant Allele frequency

Question

[Maryam]

Hello,

I wanted to say that I have analyze my NGS data and now I have a sheet of my variants along with their Allele frequencies in each of my sample (from VCF and Bam files).

I wanted to ask what threshold of Allele frequency should I set to pick my good quality variants? e.g. some people say that if Allele frequency in VCF file is more than 20% in VCF file it means your Variant is of good quality. Also what threshold people set for variant calling.

Can someone help me please.

Thank you

[4galaxy77]

Allele frequency isn't a direct measure of quality. Rare variants at less than 1% frequency do have a lower genotyping / imputation accuracy. What cutoff you use depends on what you are doing with your variants.

[Maryam]

Ok if one variant has allele frequency 40% in a sample and other variant has allele frequency 9% in the same sample. Then which variant is authentic to select?

Answer 1

[Prash]

Well said 4galaxy77. And it all depends on the phenotype of your interest! In some cases, if you were to consider MAF<0.01 You could also extend it for extremely rare variants with MAF 0.01% which would be 0.0001 which in turn depends on su-population/demographics of the disease

[Maryam]

Hello Prash,

Thank you. I am actually working on VCF allele frequencies. all the variants are showing different % in different cancer samples. Now I am confused which of them are selectable variants based on VCF and Bam allele frequencies

[Prash]

Maryam, If you did run a pipeline to call variants ( varscan, VT/Annovar, freebayes etc), you would in fact get the frequencies and if they are showing different %, that indicates the heterogeneity associated with your samples. Can you please elaborate and perhaps paste a screenshot?