Hello,
We are working with a cohort of ~800 WGS samples and have generated SV calls using Manta. For downstream analyses, we annotate each SV according to whether a matching event is observed in gnomAD-SV.
One concern is that differences between our SV discovery/filtering pipeline and the gnomAD-SV pipeline may have a larger impact on apparent presence/absence than the underlying biology. We are therefore looking for best practices to harmonize our filtering strategy with gnomAD-SV as much as possible.
At the moment, our plan is to exclude any SV for which at least one breakpoint falls within regions listed in the gnomAD depth_exclude_list or pesr_exclude_list. However, we are unsure whether this is sufficient or whether additional filtering steps would be recommended.
Would you recommend:
-
restricting analyses to specific SV size ranges
-
excluding particular genomic regions beyond the
depth_exclude_listandpesr_exclude_list(exclude_intervals_for_gcnv_filter_intervals? other?) -
any other harmonization steps?
Our main goal is to ensure that presence/absence comparisons with gnomAD-SV are not dominated by differences in calling and filtering strategies.
Thank you +++ for your advice.