Detecting multiple causal rare variants in exome sequence data

Abstract

Recent advances in sequencing technology have presented both opportunities and challenges, with limited statistical power to detect a single causal rare variant with practical sample sizes. To overcome this, the contributors to Group 1 of Genetic Analysis Workshop 17 sought to develop methods to detect the combined signal of multiple causal rare variants in a biologically meaningful way. The contributors used genes, genome location proximity, or genetic pathways as the basic unit in combining the information from multiple variants. Weaknesses of the exome sequence data and the relative strengths and weaknesses of the five approaches are discussed.

Lessons learned from Genetic Analysis Workshop 17: transitioning from genome‐wide association studies to whole‐genome statistical genetic analysis

“…When the SVs are rare, it becomes necessary to collapse rare variants into new derived variables in order to analyze them. Many different methods were used to collapse or aggregate SVs at GAW17, and these methods are summarized by König et al [2011], Melton and Pankratz [2011], Sun et al [2011], Sung et al [2011], Tintle et al [2011], and Ye and Engelman [2011]. In general, the analysis of collapsed variants had somewhat better power than that of uncollapsed variants, although the performance of collapsing and aggregating methods depended to a large extent on the underlying genetic structure and the use of unrelated individuals or family data.…”

“…The methods described by Ye and Engelman [2011] and by Tintle et al [2011], for example, illustrate a number of collapsing schemes, some previously proposed and some novel. For the most part, these methods focused on the sample of unrelated individuals.…”

“…Not unexpectedly, results from these methods identified genes containing SVs with the highest locus‐specific heritability, for example, SVs in FLT1 and KDR in trait Q1. Ye and Engelman [2011] noted that the number of SVs in the derived collapsed variants was problematic because some genes had many SVs, whereas others only had one. Taking this one step further, Cantor and Wilcox [2011] reported a variety of methods for aggregating SVs into haplotypes and multiallelic genotypes in the Group 13 summary.…”

“…These correlations may be due to chance, the quality of the sequence alignment, linkage disequilibrium, gametic disequilibrium, or other unknown structural genetic relationships. Bailey‐Wilson et al [2011], Cantor and Wilcox [2011], Sun et al [2011], Tintle et al [2011], and Ye and Engelman [2011] all noted that in the population‐based samples, the power to detect associations was not high, even for those SVs with the highest locus‐specific heritabilities (e.g., FLT1 and KDR in trait Q1), and that the type I error rate was inflated, sometimes markedly so. Tintle et al [2011] postulated that the elevation in type I error rates could be due to either population stratification or correlations between SVs on different chromosomes.…”

Considering interactive effects in the identification of influential regions with extremely rare variants via fixed bin approach

“…The possible influence of rare variants on disease susceptibility has garnered more attention in recent years[ 2 , 3 ]. A rare variant is defined by a frequency of less than 1% [ 4 ].…”