Multi-SKAT: General framework to test multiple phenotype associations of rare variants

Abstract

In genetic association analysis, a joint test of multiple distinct phenotypes can increase power to identify sets of trait-associated variants within genes or regions of interest. Existing multi-phenotype tests for rare variants make specific assumptions about the patterns of association of underlying causal variants, and the violation of these assumptions can reduce power to detect association. Here we develop a general framework for testing pleiotropic effects of rare variants based on multivariate kernel regression (Multi-SKAT). Multi-SKAT models effect sizes of variants on the phenotypes through a kernel matrix and performs a variance component test of association. We show that many existing tests are equivalent to specific choices of kernel matrices with the Multi-SKAT framework. To increase power to detect association across tests with different kernel matrices, we developed a fast and accurate approximation of the significance of the minimum observed p-value across tests. To account for related individuals, our framework uses a random effects for the kinship matrix. Using simulated data and amino acid and exomearray data from the METSIM study, we show that Multi-SKAT can improve power over single-phenotype SKAT-O test and existing multiple phenotype tests, while maintaining type I error rate.

Subset-Based Analysis Using Gene-Environment Interactions for Discovery of Genetic Associations across Multiple Studies or Phenotypes

“…For example, Dutta et al [37] have studied the effect of modeling kernel matrix in the score test statistic using estimated phenotype covariance for testing pleiotropic effects of rare variants based on multiple kernel regression (Multi-SKAT). While we only use phenotypic correlation for characterizing the distribution of our combined statistics instead of using it directly during the process of constructing the statistics, incorporating phenotypic correlation into the test statistics may further improve the power of our method.…”

Fast and flexible linear mixed models for genome-wide genetics

“…Linear mixed effect models (LMMs) are widely used to account for 9 non-independent samples in quantitative genetics [9]. The flexibility and interpretability 10 of LMMs make them a dominant statistical tool in much of biological research [9][10][11][12][13][14][15][16][17][18].…”

“…General-purpose tools for this are too slow to be 14 practical for genome-scale datasets with thousands of observations and millions of 15 genetic markers [19]. This lack of scalability is caused primarily by two factors: (i) 16 closed-form solutions of maximum-likelihood (ML or REML) or posterior estimates of 17 the variance components are not available and numerical optimization routines require 18 repeatedly evaluating the likelihood function many times, and (ii) each evaluation of the 19 likelihood requires inverting the covariance matrix of random effects, an operation that 20 scales cubically with the number of observations. Repeating this whole process millions 21 of times quickly becomes infeasible.…”