Going to extremes to improve power - Impact of Extreme Phenotype Sampling on population stratification

Speaker: Kelly Burkett (uOttawa)

Time: 1:30-2:30 pm, Mar 31 2017

Room: KED B015

Abstract:

Extreme Phenotype Sampling (EPS) has been proposed as a cost-effective strategy for genetic association studies when the cost of collecting genotype data is high. EPS involves sampling individuals in the upper and lower percentiles of the phenotype distribution as the phenotypic extremes are expected to be enriched for causal genetic variants. This design has been shown to have higher power; however, little attention has been given to the effects of confounding due to population stratification under an EPS design.

In this presentation, I will first show how the false positive rate due to confounding is increased under the EPS design when compared to a random sample of equivalent size. I will then describe the performance under EPS of a common principal component-based approach to correcting for population stratification.