Genome-wide mapping of quantitative trait loci in admixed populations using mixed linear model and Bayesian multiple regression analysis

Abstract

BackgroundPopulation stratification and cryptic relationships have been the main sources of excessive false-positives and false-negatives in population-based association studies. Many methods have been developed to model these confounding factors and minimize their impact on the results of genome-wide association studies. In most of these methods, a two-stage approach is applied where: (1) methods are used to determine if there is a population structure in the sample dataset and (2) the effects of population structure are corrected either by modeling it or by running a separate analysis within each sub-population. The objective of this study was to evaluate the impact of population structure on the accuracy and power of genome-wide association studies using a Bayesian multiple regression method.MethodsWe conducted a genome-wide association study in a stochastically simulated admixed population. The genome was composed of six chromosomes, each with 1000 markers. Fifteen segregating quantitative trait loci contributed to the genetic variation of a quantitative trait with heritability of 0.30. The impact of genetic relationships and breed composition (BC) on three analysis methods were evaluated: single marker simple regression (SMR), single marker mixed linear model (MLM) and Bayesian multiple-regression analysis (BMR). Each method was fitted with and without BC. Accuracy, power, false-positive rate and the positive predictive value of each method were calculated and used for comparison.ResultsSMR and BMR, both without BC, were ranked as the worst and the best performing approaches, respectively. Our results showed that, while explicit modeling of genetic relationships and BC is essential for models SMR and MLM, BMR can disregard them and yet result in a higher power without compromising its false-positive rate.ConclusionsThis study showed that the Bayesian multiple-regression analysis is robust to population structure and to relationships among study subjects and performs better than a single marker mixed linear model approach.

Statistical power in genome-wide association studies and quantitative trait locus mapping

“…Assuming that we do not want to claim loci associated with population structure as significant in GWAS and fitting population structure is necessary, this study is the first to theoretically evaluate the effects of population structures on the statistical powers. The conclusions are consistent with the empirical observations from simulation studies (Toosi et al 2018). However, if population structural effects are present but ignored in the mixed model, the statistical power will be reduced compared to that if they are taken into account (see Fig.…”

“…In addition to cryptic relatedness, population structure is another factor that needs to be controlled in GWAS (Pritchard et al 2000b). Effects of population structure on the powers of GWAS have been investigated via Monte Carlo simulations (Atwell et al 2010;Platt et al 2010;Korte and Farlow 2013;Shin and Lee 2015;Toosi et al 2018). A consensus conclusion is that proper control of population structure can reduce false positive rate.…”

“…Population structure is often caused by population heterogeneity (or admixture) represented by multiple ethnic groups or subpopulations within the association population (Pritchard et al 2000a(Pritchard et al , 2000b. The purpose of fitting population structure effects into the LMM is to reduce false positives for loci that are confounded with population structure (Toosi et al 2018). For example, if a locus is fixed to alleles unique to subpopulations and the subpopulations are strongly associated with the trait under study, we do not want to claim the locus as associated with the trait because the association may be caused by subpopulations.…”

Assessing Accuracy of Genomic Predictions for Resistance to Infectious Hematopoietic Necrosis Virus With Progeny Testing of Selection Candidates in a Commercial Rainbow Trout Breeding Population

“…The wssGBLUP mimics the Bayesian variable selection model by fitting all SNPs in the multiple regression model but assigning differential weights to the SNPs based on the variance of each SNP effect (8,24). Lately, Bayesian variable selection models that use a single-step approach have been developed (25)(26)(27)(28), including the single-step Bayesian multiple regression (ssBMR) method (8,25,28).…”

A genomewide association study in divergently selected lines in rabbits reveals novel genomic regions associated with litter size traits

“…Together with these tools, several methods for genomewide association analyses have been also already developed and applied in many different species (Fan, Du, Gorbach, & Rothschild, 2010). Among them, genomewide association studies (GWAS) using multimarker regression approaches can attain better power detection to identify genomic regions associated with a trait than the classical approach of single maker simple regression (López de Maturana et al, 2014;Toosi, Fernando, & Dekkers, 2018).…”