High-depth filter for Mutect calls
We are working on a cancer genome project and we use Mutect to call somatic mutations.
We have a question related to the high-coverage depth filtering. Below is a list of nearby positions output by Mutect
where the coverage in the tumor sample is very high but the coverage in the normal sample is not.
Partial output from *call_stats.out file -
contig position t_q20_count n_q20_count failure_reasons judgement
2 89849000 622 108 KEEP
2 89850117 508 152 KEEP
2 89850498 713 732 KEEP
2 89850649 583 403 KEEP
2 89850993 849 142 KEEP
2 89872002 540 286 KEEP
2 89877959 607 259 KEEP
These positions are not filtered out by Mutect.
I wonder why is a high-depth filter not used?
Would it make sense to filter these variants and which high-depth threshold should we choose?