Closed bredelings closed 3 years ago
To be explicit, the AD counts are 1163,135, and 30. So the allele frequencies are approximate 0.88, 0.1, and 0.02. Clearly 0.02 is small (<0.05), but 0.1 should be fine.
This is as much to do with the VCF spec as anything else - there's no specified way to filter by alleles or haplotypes. You can use some kind of custom filter naming that refers to each allele (e.g. AF_0
, AF_1
, ... etc), but no tools will know how to use this. Actually, the spec if pretty vague on what a FILTER
'rd record actually means, in particular when there are multiple samples. Octopus also uses FORMAT/FT
, and the interpretation is "the called genotype cannot be trusted", so if any of the called alleles look dodgy then the genotype will be filtered.
This is as much to do with the VCF spec as anything else
Well, I am not shocked.
Octopus also uses FORMAT/FT, and the interpretation is "the called genotype cannot be trusted", so if any of the called alleles look dodgy then the genotype will be filtered.
Good to know!
Thanks for all the help.
Describe the bug
VCF lines are being flagged as AF instead of PASS because one allele has a low frequency, when there is a second ALT allele that has a high frequency.
Version
Example
So here is a record from running polyclone with
--max-clones 5
. This line:gets flagged by the AF < 0.05 filter. However, only one of the two ALT alleles has AF < 0.05.
If there is only one alt allele, then censoring the allele and censoring the line are then same, but here that seems not to be the case. Thoughts?