Methods for detecting associations between phenotype and aggregations of rare variants

Although genome-wide association studies have uncovered variants associated with more than 150 traits, the percentage of phenotypic variation explained by these associations remains small. This has led to the search for the dark matter that explains this missing genetic component of heritability. One potential explanation for dark matter is rare variants, and several statistics have been devised to detect associations resulting from aggregations of rare variants in relatively short regions of interest, such as candidate genes. In this paper we investigate the feasibility of extending this approach in an agnostic way, in which we consider all variants within a much broader region of interest, such as an entire chromosome or even the entire exome. Our method searches for subsets of variant sites using either Markov chain Monte Carlo or genetic algorithms. The analysis was performed with knowledge of the Genetic Analysis Workshop 17 answers.