Collapsing approaches for the association analysis of rare variants

Abstract

In testing for associations with rare variants, alternative methods to those used for common SNPs are required due to their lack of power at lower frequency. Collapsing approaches overcome this power loss by testing for an association with an aggregate of rare variants. These tests pool information across the rare variants such that a single test is performed on the summary statistic, and are powerful tools, provided that certain conditions are satisfied. In this chapter, the general framework of collapsing methods is explored, including optimal conditions for attaining high power. Comparisons are made between specific collapsing methods, as well as data-adaptive versions that have been developed to recover much of the power loss from nonideal settings.