Performance Gains in Genome-Wide Association Studies for Longitudinal Traits via Modeling Time-varied effects

Chao Ning,Lei Zhou,Jinluan Fu,Shengli Zhang,Haifei Wang,Huimin Kang,Jianfeng Liu,Aiguo Wang,Dan Wang

doi:10.1038/s41598-017-00638-2

Abstract

Complex traits with multiple phenotypic values changing over time are called longitudinal traits. In traditional genome-wide association studies (GWAS) for longitudinal traits, a combined/averaged estimated breeding value (EBV) or deregressed proof (DRP) instead of multiple phenotypic measurements per se for each individual was frequently treated as response variable in statistical model. This can result in power losses or even inflate false positive rates (FPRs) in the detection due to failure of exploring time-dependent relationship among measurements. Aiming at overcoming such limitation, we developed two random regression-based models for functional GWAS on longitudinal traits, which could directly use original time-dependent records as response variable and fit the time-varied Quantitative Trait Nucleotide (QTN) effect. Simulation studies showed that our methods could control the FPRs and increase statistical powers in detecting QTN in comparison with traditional methods where EBVs, DRPs or estimated residuals were considered as response variables. Besides, our proposed models also achieved reliable powers in gene detection when implementing into two real datasets, a Chinese Holstein Cattle data and the Genetic Analysis Workshop 18 data. Our study herein offers an optimal way to enhance the power of gene detection and further understand genetic control of developmental processes for complex longitudinal traits.

Highlights

In previous quantitative trait loci (QTL) linkage analysis on longitudinal traits, three statistical strategies are proposed as follows: The first one is based on repeatability model or multivariate model, which treats the multi-point measured trait as repeated measurements of the same trait or as different traits[9, 10]
A series of simulation studies were performed to investigate the properties of the proposed models, and to compare with previous developed models, i.e., genome-wide association studies where estimated breeding value (EBV) or deregressed proof (DRP) were used as response variable with polygenic effects modelled (GWAS-EBV-P or GWAS-DRP-P), genome-wide association studies where EBVs or DRPs were used as response variable without polygenic effects modelled (GWAS-EBV-NP or GWAS-DRP-NP), and genome-wide association studies where estimated residuals were used as response variable (GWAS-Residual)
As the false positive rates (FPRs) were independent of the Quantitative Trait Nucleotide (QTN) heritability in the simulation, we averaged the FPRs across different QTN heritabilities

Summary

Introduction

In previous quantitative trait loci (QTL) linkage analysis on longitudinal traits, three statistical strategies are proposed as follows: The first one is based on repeatability model or multivariate model, which treats the multi-point measured trait as repeated measurements of the same trait or as different traits[9, 10]. The third one is based on varying coefficient model, which fits the coefficients of genetic and environmental effects as the linear regression on a set of splines or polynomials of time to model the time-varied effects[14,15,16,17] Varying coefficient model is only suitable to well-structured dada where all individuals must be measured at the fixed time points The drawbacks for these strategies limit their further application in the GWAS. Random regression model was suitable for QTL detecting in the presence of gene by environment interactions[31] Application of this model in GWAS has not been fully surveyed so far. We further validated our model with a Chinese Holstein cattle data and the Genetic Analysis Workshop 18 (GAW18)[32]

Methods

Results

Conclusion