Early candidate-gene sequencing studies for complex traits were based on the comparison of numbers of non-synonymous alleles exclusive to cases or controls (or samples at the extremes of the trait distribution)21, 26. This approach has limited power because it ignores common and low-frequency polymorphisms, as most such variants would be present in cases and controls. Recently, a number of statistical tests have been designed for rare-variant analysis. The Combined Multivariate and Collapsing (CMC) test58 jointly assesses the role of rare and common variation. For the common variants, traditional regression-based association is applied. For rare variation, an individual's predictor in a regression model is defined as 1 if the individual possesses at least one rare variant in the region (e.g., gene) and 0 otherwise. The Weighted-Sum Statistic (WSS) test59, creates a composite genotype score for all individuals. This score is the sum of alternate alleles weighted by the inverse of the binomial variance. A rank sum test is then performed on the genotype scores between phenotypic groups. The Kernel-Based Adaptive Cluster (KBAC) test60 also uses a weighting scheme that reflects apparent
