luntergroup / octopus

Bayesian haplotype-based mutation calling
MIT License
302 stars 38 forks source link

gVCF? #107

Closed ghost closed 4 years ago

ghost commented 4 years ago

Hello,

I was wondering if support for the gVCF format was planned? I also notice there is no VAF field in the output. thanks

dancooke commented 4 years ago

Reference calling is available for the individual calling model. See the --refcall command line option.

You might want to checkout the feature/discontiguous-phasing branch as this contains some performance improvements related to reference calling - probably this will be merged into develop soon.

ghost commented 4 years ago

Ok thanks, and regarding VAF it field it's a design choice not to have it? I stick with this version because I don't have the forest files otherwise.

dancooke commented 4 years ago

INFO VAF or FORMAT VAF?

ghost commented 4 years ago

I guess FORMAT, as I mean at the level of my sample

dancooke commented 4 years ago

You can use --annotations VAF to get VAF annotations during filtering (so long as your filter uses the VAF measure).

ghost commented 4 years ago

You can use --annotations VAF to get VAF annotations during filtering (so long as your filter uses the VAF measure).

I am not sure to understand how to enforce VAF in the filters as I don't see it in the user manual.

./octopus-0.6.3-beta/bin/octopus -R shasta_final.fa -I ARCcestor.sorted.bam -o ARCcestor.octopus.vcf.gz --forest 0.6.3-Forest/germline.v0.6.3-beta.forest --threads 24 -k ARCcestor.bed --refcall --annotations VAF

This is my command but obviously there is no VAF output. Similarly, how to get QUAL? I see in the manual: --filter-expression "QUAL < 10"

But let's say I just want the information, because I don't know beforehand what threshold to apply, then one needs to do something like --filter-expression "QUAL, VAF", etc ?

And what's the difference with the DENOVO filter? Aren't all candidates "Denovo" per default (or is this option used only when one has a pedigree available)?

Sorry for the many questions.