Gwas: Adjust For Population Stratification In Impute2 Output

Question

[w0nderingatwork]

Dear all,

I'm working with a GWAS dataset and I am not so experienced yet with imputation of missing genotypes. But I managed to phase my dataset with SHAPEIT, impute with IMPUTE2 and did an association analysis with SNPTEST, so far so good. But: how do I now adjust for population stratification? I don't find any option in SNPTEST to adjust for Genomic Control, so I guess it can be done by adjusting for principal components as covariates. But how do I calculate principal components from IMPUTE2 output (file types: .gen; .gen_info)??

Thanks a lot! Frank

[Maxime Lamontagne]

You can transform IMPUTE2 output with gtool. http://www.well.ox.ac.uk/~cfreeman/software/gwas/gtool.html

"GTOOL is a program for transforming sets of genotype data for use with the programs SNPTEST and IMPUTE. "

[w0nderingatwork]

Hi Maxime, thanks a lot, that is a good option. The only drawback is that I loose the probability values for the imputed genotypes, because GTOOL just applies a propability threshold and will then convert the imputed genotypes to plink format such as regularly typed genotypes. That means that in subsequent association analysis in plink the uncertainty of the imputed genotypes is not considered any more. Best regards, Frank

[zx8754]

Calculate principle components on genotyped set, then use it for imputed set, see if it helps to reduce the inflation.

[w0nderingatwork]

thank you! I will try that. Best regards

Answer 1

[kumar.vinod81]

Hi, I dont know much about IMPUTE2 for population stratification but after converting your data in HapMap, Plink or flapjack format, you can import data in either TASSEL or GAPIT or in STRUCTURE, where you can run them for PCA analysis or structure analysis. Thanks,

Vinod