Impact of missing genotype imputation on the power of Genome Wide Association Studies.

Genome Wide Association Studies are often performed on datasets containing a relatively small number of genotypes. In those cases, statistical tests performed on subgroups of those genotypes lack power or may even not fulfill the requirements of minimal number of observations. In this work we present results of running a GWAS analysis including parametric and nonpara- metric analysis of variance for different stress markers and models, validation by a group of related patients and clustering of results by their chromosomal positions. We show that the selection of imputation method has a significant impact at each phase of the analysis, and that it is worth to use the best method available.