A variational Bayes discrete mixture test for rare variant association.

摘要: Recently, many statistical methods have been proposed to test for associations between rare genetic variants and complex traits. Most of these association by aggregating variations within a predefined region, such as gene. Although there is evidence that "aggregate" tests are more powerful than the single marker test, generally ignore neutral therefore unable identify specific driving with phenotype. We propose novel aggregate rare-variant explicitly models fraction neutral, at gene-level, infers rare-variants association. Simulations show in practical scenario where given region genome only causal our approach has greater power compared other popular Sequence Kernel Association Test (SKAT), Weighted Sum Statistic (WSS), collapsing method Morris Zeggini (MZ). Our algorithm leverages fast variational Bayes approximate inference methodology scale exome-wide analyses, significant computational advantage over exact model selection methodologies. To demonstrate efficacy we von Willebrand Factor (VWF) levels VWF missense imputed from National Heart, Lung, Blood Institute's Exome Sequencing project into 2,487 African Americans suggests relatively small (~10%) strongly associated lower Americans.