SNV occuring 2 fold more in forward than in reverse reads should be filtered?

Question

[jeni]

Hi!

I am reviewing the SNVs of a vcf in IGV. I have seen that some of them occurs almost 2 fold in forward than in reverse reads (or viceversa) of a gene panel deep sequencing bam.

As an example: I have an SNV (G>T) supported by 795 forward reads and 496 reverse reads. The reference allele is supported, however, by 6677 forward reads and 6716 reverse reads. I know that a clear bias by strand is indicative of a sequencing error, so I was wondering if I should consider as a bias the support of the variant allele, and therefore, if I should filter it.

Thanks!