Phasing And Imputation Using A Combination Of Pedigrees And Population Data

Question

[Lrk]

Through 23andMe, I have SNP genotypes from family members spanning three generations, incl. one parent and a sibling of the missing parent, multiple siblings in the middle generation, one child in the the third generation and the child's other parent, and some cousins. Some of these have around 500,000 markers, and some of them have around a million.

I want to phase these and infer the genotypes of the missing parent, and also use imputation to bring each individual as close to the million snps as reliably possible, using a combination of information contained within the pedigree and also data from sources like the 1000 Genomes Project. Do you have any advice on how to accomplish this or which programs if any would work?

Answer 1

[Michael]

List of tools and documentation (there are more ofc, could someone link review papers on imputation and phasing?):

For phasing: e.g. fastPhase

For imputation:

• IMPUTE
• MACH
• BimBam

Answer 2

[Zev.Kronenberg]

I like BEAGLE it is fast and you get Identity by decent from it as well. BEAGLE LINK

I would be careful about how much data you are missing. I don't allow a loci to have more than 5% missing data for all individuals.