Covariate Selection for Association Screening in Multi-Phenotype Genetic studies

Testing for associations in big data faces the problem of multiple comparisons, with true signals difficult to detect on the background of all associations queried. This is particularly true in human genetic association studies where phenotypic variation is often driven by numerous variants of small effect. The current strategy to improve power to identify these weak associations consists of applying standard marginal statistical approaches and increasing study sample sizes. While successful, this approach does not leverage the environmental and genetic factors shared between the multiple phenotypes collected in contemporary cohorts. Here we develop Covariates for Multi-phenotype Studies, an approach that improves power when correlated variables have been measured on the same samples. Our analyses over real and simulated data provide direct support that correlated phenotypes can be leveraged to achieve dramatic increases in power, often surpassing the power equivalent of a two-fold increase in sample size.