imputation and changing genotypes in Beagle and FImpute

Question

[Wietje]

Hello fellow Biostars,

I am currently aiming to impute genotypes from different SNP-Chip resolutions (6k, 50k, 777k...) and have tried Beagle 4.0 and FImpute because I want to include pedigree information.

The process seems successful so far, however, FImpute notifies me that changes to the original genotypes have been made (due to mendelian errors I presume). As far as I know there is no way to prevent the programme from doing so.

Beagle on the other hand never gave a warning or other that genotypes have been altered - I tried checking and have the feeling that it doesn't happen.

Do you know if Beagle (4.0) also changes genotypes during imputation if it encounters pedigree/ mendelian errors?

Thanks, Wietje

Answer 1

[Shab86]

If you want to utilize pedigree information for imputation later on, why not use the SHAPEIT's duoHMM tool for better pre-phasing of your genotyped datasets. Then you could go for imputation using either FImpute or Beagle. Doing this would reduce computational costs and errors too.

[Wietje]

Thank you for the suggestion, I will look into it for sure - unfortunately, this does not answer my initial question though...

[Shab86]

I know it doesn't but I want to use duoHMM first and then use FImpute vs FImpute alone to check whether any mendel errors are prevented or not and if yes, then by how much. Beagle though quite fast won't give you genotype changed information.