Closed xunjieli closed 5 years ago
:point_down: :man_facepalming: :point_down:
https://github.com/dnanexus-rnd/GLnexus/blob/master/src/discovery.cc#L71-L72
I think a possible fix will be to have merge_discovered_alleles
recognize when the same deletion/MNP is being observed in multiple discovery ranges based on the discovered_allele_info::in_target
field. Since the latter is optional, I need to double check that the latter field will be populated as expected in all cases.
Thanks, mlin@. The fix makes sense to me.
@xunjieli I implemented that fix on master, pls reopen if you still hit related issues. Expecting to cut a new point release next week after some other wip lands. Thanks!
Thanks @mlin. I've verified that this commit fixes the issue that I saw.
When exome capture regions is used as --bed, there are issues with AF calculation at the edges of these regions.
Here's one example where GLnexus will calculate the AF as 1.33333 when it should be 0.666667
HG002
HG003
HG004
Bed file used (taken from agilent_sureselect_human_all_exon_v5_b37_targets.bed)
This merged.bcf will have
AF=1.33333
:If repeat the above, but supply
19 0 10000000
as --bed, GLnexus is correct in the AF calculation.AF=0.666667
: