IMPUTE2 variable question

Question

[raalsuwaidi]

Dear all

I am trying to perform an imputation on a VCF file that I have but using my own reference panel. I am getting confused with the parameters though. Here is what I understand:

-g is my vcf file that I am trying to impute 

-known_haps_g is the hap file of the file I am planning to impute in case I do not provide it in -g

-m is the recombination file as per the sample provided by impute2

-h the haplotypes for the reference file

-l the legend file of the reference file

Is that correct?

Answer 1

[Kevin Blighe]

You have generally got it correct, however, -known_haps_g is used in comjunction with -use_prephased_g, and is the output of a pre-phasing step, ie., with SHAPEIT2. If you use these parameters, then you do not have to supply -g. If you use -g, the file should be in gens format.

All explanations can be found here: https://mathgen.stats.ox.ac.uk/impute/impute_v2.html

If you want a workflow for pre-phasing with SHAPEIT2, then I have coincidentally provided an answer that covers this:

• C: Phasing with SHAPEIT

Another related answer:

• A: Steps for genotype imputation?

SHAPEIT2 can accept PLINK data as input. You can convert VCF to PLINK with PLINK, if you want.

Kevin

[raalsuwaidi]

Thank you so much. But how can I change my vcf to gen formate

[Kevin Blighe]

Please take a look at an answer from the PLINK developer: C: How to convert vcf to impute2 format?

Alternatively, if you want to go the pre-phasing route, you can import / convert your VCF to PLINK format via PLINK itself (--vcf or --bcf flags), and then use that as input to SHAPEIT2. SHAPEIT2's output then feeds directly into IMPUTE2 via -use_prephased_g -known_haps_g