Identifying rare variants for quantitative traits in extreme samples of population via Kullback-Leibler distance

Xinrong Xiang,Yang Xiang,Yumei Li

doi:10.1186/s12863-020-00951-2

Xinrong Xiang, Yang Xiang + Show 1 more

Open Access

https://doi.org/10.1186/s12863-020-00951-2

Copy DOI

Journal: BMC Genetics	Publication Date: Nov 24, 2020
Citations: 1	License type: open-access

Affiliation: Hunan Normal University, Huaihua University

Abstract

BackgroundThe rapid development of sequencing technology and simultaneously the availability of large quantities of sequence data has facilitated the identification of rare variant associated with quantitative traits. However, existing statistical methods depend on certain assumptions and thus lacking uniform power. The present study focuses on mapping rare variant associated with quantitative traits.ResultsIn the present study, we proposed a two-stage strategy to identify rare variant of quantitative traits using phenotype extreme selection design and Kullback-Leibler distance, where the first stage was association analysis and the second stage was fine mapping. We presented a statistic and a linkage disequilibrium measure for the first stage and the second stage, respectively. Theory analysis and simulation study showed that (1) the power of the proposed statistic for association analysis increased with the stringency of the sample selection and was affected slightly by non-causal variants and opposite effect variants, (2) the statistic here achieved higher power than three commonly used methods, and (3) the linkage disequilibrium measure for fine mapping was independent of the frequencies of non-causal variants and simply dependent on the frequencies of causal variants.ConclusionsWe conclude that the two-stage strategy here can be used effectively to mapping rare variant associated with quantitative traits.

Highlights

The rapid development of sequencing technology and simultaneously the availability of large quantities of sequence data has facilitated the identification of rare variant associated with quantitative traits
Rare variant association analysis in the past several years mainly focused on the qualitative trait
We investigate the performance of the proposed method and compare it with three commonly used methods of the burden test [12], the sequence kernel association test (SKAT) [17], and the optimal test that combines SKAT and the burden test (SKAT-O) [10]

Summary

Introduction

The rapid development of sequencing technology and simultaneously the availability of large quantities of sequence data has facilitated the identification of rare variant associated with quantitative traits. One approach for rare variant association analysis of quantitative traits is the linear regression model. Most regression-based methods rely on the normality assumption of the phenotype [8, 21]. Another commonly used approach adopts phenotype extreme selection design where one can transform the quantitative trait association study into case-control association study of qualitative traits by treating the upper extreme as cases and the lower extreme as controls in a strategy using extreme phenotype [22,23,24,25]. A smaller sample size for extreme-phenotype sampling than that for random sampling is needed to achieve similar power [23, 24]

Methods

Results

Discussion

Conclusion