A biologically informed method for detecting associations with rare variants

Abstract

With the recent flood of genome sequence data, there has been increasing interest in rare variants and methods to detect their association to disease. Many of these methods are collapsing strategies which bin rare variants based on allele frequency and functional predictions; but at this point, most have been limited to candidate gene studies with a small number of candidate genes. We propose a novel method to collapse rare variants based on incorporating biological information from the public domain. This paper introduces the functionality of BioBin, a biologically informed method to collapse rare variants and detect associations with a particular phenotype. We tested BioBin using low coverage data from the 1000 Genomes Project and discovered appropriate binning characteristics based on what one might expect given the size of the gene. We also tested BioBin using the pilot targeted exome data from 1000 Genomes Project. We used biologically-informed binning and differences in minor allele frequencies as a means to distinguish between two ancestral populations. Although BioBin is still in developmental stages, it will be a useful tool in analyzing sequence data and uncovering novel associations with complex disease. © 2012 Springer-Verlag.