Convert VCF to BGEN 1.3

Question

[Vince]

I have imputed genotype data from the Haplotyle Reference Consortium in VCF format, and would like to convert to BGEN 1.3 (or 1.2). I know that qctool version 2 support this, as does plink version 2.0.

I would like to inquire as to experiences anyone has had converting genotype probabilities (GP field in VCF) using these tools:

1. Is one tool preferred over the other?
2. Any caveats we should be aware of?

I would tend to use plink, as it is quite popular and actively developed, although the developers of qctool also developed the BGEN format, which may make it more ideal. I would assume these tools are equivalent for the task of converting VCF to BGEN, but your experience with this may suggest otherwise.

Answer 1

[chrchang523]

plink2 supports allelic dosages, not genotype probabilities. So if your VCF GP field says P(0/0)=0.2, P(0/1)=0.52, P(1/1)=0.28, plink2 will just save "dosage(1)=1.08", and then when it's asked to export to BGEN it'll encode the dosage as P(0/0)=0, P(0/1)=0.92, P(1/1)=0.08.

When this is fine, plink2 is a good choice. When this isn't good enough, use something like qctool.

[Vince]

Do you have link to the plink website where this is stated? This would he helpful.

[Vince]

Found it here, https://www.cog-genomics.org/plink/2.0/input, "Dosage import settings" section.