Longitudinal data analysis for rare variants detection with penalized quadratic inference function

Hongyan Cao,Zhi Li,Yuehua Cui,Yanbo Zhang,Haitao Yang

doi:10.1038/s41598-017-00712-9

Abstract

Longitudinal genetic data provide more information regarding genetic effects over time compared with cross-sectional data. Coupled with next-generation sequencing technologies, it becomes reality to identify important genes containing both rare and common variants in a longitudinal design. In this work, we adopted a weighted sum statistic (WSS) to collapse multiple variants in a gene region to form a gene score. When multiple genes in a pathway were considered together, a penalized longitudinal model under the quadratic inference function (QIF) framework was applied for efficient gene selection. We evaluated the estimation accuracy and model selection performance under different model settings, then applied the method to a real dataset from the Genetic Analysis Workshop 18 (GAW18). Compared with the unpenalized QIF method, the penalized QIF (pQIF) method achieved better estimation accuracy and higher selection efficiency. The pQIF remained optimal even when the working correlation structure was mis-specified. The real data analysis identified one important gene, angiotensin II receptor type 1 (AGTR1), in the Ca2+/AT-IIR/α-AR signaling pathway. The estimated effect implied that AGTR1 may have a protective effect for hypertension. Our pQIF method provides a general tool for longitudinal sequencing studies involving large numbers of genetic variants.

Highlights

Longitudinal data are often observed in biomedical studies with repeated measures of the same subject over time
Methods for detecting rare variants have been developed and can be broadly classified into three categories: (1) burden tests, for example, the weighted sum statistic (WSS) methods[9]; (2) variance component-based tests represented by the sequence kernel association test (SKAT)[10]; and (3) dimension-reduction based tests such as functional principal components analysis (FPCA)[11] and the adaptive ridge regression method[12]
We explored gene-based association studies for next-generation sequencing data with longitudinal measures of binary phenotypic traits using the penalized QIF (pQIF) method

Summary

Introduction

Longitudinal data are often observed in biomedical studies with repeated measures of the same subject over time. Very few methods have been developed or extended to detect rare variants associated with longitudinal disease traits[13,14,15,16]. Wu et al.[14] and Chiu et al.[13] summarized the rare variants longitudinal studies, where most of the statistical models were based on GEE and LM models These methods face computational challenges with limited sample size and missing data. The classical methods faces estimation instability issues when the number of variants is large This motivates us to adopt a penalized regression method for better parameter estimation, and achieving gene selection in the mean time. When a large number of gene variables are modelled simultaneously in a regression model, high-dimensional variable selection strategies become essential for a genetic association study. Penalized regression methods have been applied to rare variants association analysis when a univariate disease trait is considered[22,23,24]

Objectives

Methods

Findings

Conclusion