FAQ

Algorithms for phenotype imputation

Q. Are there any restrictions on which machine learning algorithms can be used for phenotype imputation?

A. No, there are no restrictions. You can use ANY algorithm you like, because the validity of POP-GWAS does not depend on the choice of algorithm. In general, we recommend using the algorithm with higher imputation accuracy (quantified by the residual correlation of observed and imputed phenotype after adjustment for GWAS covariates) to improve the power of ML-assisted GWAS.