Feature request: option to ignore filters about somalier

[brentp]

this is now in master and i'll make a new release shortly
you can do, e.g.

export SOMALIER_ALLOWED_FILTERS=LowGQX,NoPassedVariantGTs
somalier extract ...

[holtjma]

Relatedly, is there a description of the minimum requirements for a variant to be extracted? All I could find was the AD check defaulting to GT if AD isn't available.

[brentp]

i will document the requirements. briefly it requires the FILTER is one of "PASS", "", ".", "RefCall"
then, when genotyping, it uses allele-balance AB. with HOM-REF < 0.04, HOM-ALT > 0.96, HET 0.2 to 0.8 and all else is unknown.

what is the FILTER of the variants you'd like to include? I'd rather add more things like "RefCall" than add extra flags.

[holtjma]

To be honest, it's a lot of weird filters that shouldn't pass under normal conditions (things like LowGQX;NoPassedVariantGTs), but due to the way this test captures the sequencing it throws off the filters even when the variant call is good.

Sounds like my best path forward would be to just pre-parse the gVCFs such that somalier will accept the variants (i.e. reduce it down to PASS and a GT field only).

[brentp]

I'll make this an environment variable, something like SOMALIER_ALLOWED_FILTERS in a future release, for now, yes, you can use some simple awk to pre-process.