Search for compound heterozygous effects in exome sequence of unrelated subjects

To enable the assessment of compound heterozygosity, we propose a simple approach for incorporating genotype phase in a rare variant collapsing procedure for the analysis of DNA sequence data. When multiple variants are identified within a gene, knowing the phase of each variant may provide additional statistical power to detect associations with phenotypes that follow a recessive or additive inheritance pattern. We begin by phasing all marker data; then, we collapse nonsynonymous single-nucleotide polymorphisms within genes on each phased haplotype, resulting in a single diploid genotype for each gene, which represents whether one or both haplotypes carry a nonsynonymous variant allele. A recessive or additive association test can then be used to assess the relationship between the collapsed genotype and the phenotype of interest. We apply this approach to the unrelated individuals data from Genetic Analysis Workshop 17 and compare the results of the additive test with a dominant test in which phase is not informative. Analysis of the first phenotype replicate shows that the FLT1 gene is significantly associated with both Q1 and the binary affection status phenotype. This association was detected by both the additive and dominant tests, although the additive phase-informed test resulted in a smaller p-value. No false-positive results were detected in the first phenotype replicate. Analysis of the average values of all phenotype replicates correctly identified five other genes important to the simulation, but with an increase in false-positive rates. The accuracy of our method is contingent on correct phase determination.