mutect2 error in strand bias count?
I have some examples showing something wrong in calculating ALT_F1R2 and ALT_F2R1 in MuTect2. In this example, the alignment in the locally re-assemblied BAM (by MuTect2) shows that all 12 reads are aligned one directional (reverse) but ALT_F1R2 and ALT_F2R1 counts in the vcf file are 9 and 3, respectively. This SNP is actually rejected by MuTect 1 due to 'strand_artifact' which looks right but passed in MuTect2. Is this a bug in the MuTect2?