search

gusevlab / fusion_twas

methods for functional summary-based imputation
http://gusevlab.org/projects/fusion/
GNU General Public License v3.0
75 stars 45 forks source link

Add covariates when computing weigths #39

Closed PierreSolomon closed 3 years ago

PierreSolomon commented 3 years ago

Hello,

Concerning the covariates, what is the best way to deal with missing value when adding covariates with the --covar flag? Do we need to use the .cov format from plink with 2+C columns?

Thank you in advance. Best

Pierre

sashagusev commented 3 years ago

Hi Pierre, yes the format is *.cov from plink with 2+C columns. Currently FUSION does not do anything about missing data and assumes the covariates are complete. In general, I think the commonly used options are: (a) remove the individuals with missing covariates from your analysis (you can do this with plink --remove before you pass the data to FUSION), (b) fill in the missing covariates with the covariate mean (you can do this externally in R/python prior to passing the covariates to FUSION), (c) use some fancy covariate imputation method. Hope that helps!