Genome-wide association studies have problems due to confounding: Are family-based designs the answer?

Abstract

Genome-wide association studies (GWASs) can be affected by confounding.Family-based GWAS uses random, within-family genetic variation to avoid this.A study in PLOS Biology details how different sources of confounding affect GWAS and whether family-based designs offer a solution.Since the advent of large-scale, genome-wide genotype data, one study design has dominated human genetics: the genome-wide association study (GWAS).A GWAS scans across the genome for loci that are associated with a phenotype.However, GWASs are susceptible to confounding due to gene-environment correlation (environmental confounding) and correlations with other genetic variants across the genome (genetic confounding).Although GWASs use techniques including principal component analysis (PCA) to adjust for population structure, this often fails to eliminate all confounding [1-3].The confounding present in GWAS has caused issues in downstream applications, including: estimation of heritability and genetic correlation [4,5], estimation of disease causes using mendelian randomization [6], and inferences of natural selection [2].Family-based GWAS (FGWAS) has been proposed as a solution to confounding because genotypes of offspring are randomly assigned given the genotypes of the parents, generating a natural experiment.Recent FGWASs have demonstrated confounding in GWASs of several phenotypes [7,8], including educational attainment, cognitive ability, height, smoking, and age when first giving birth.In this issue of PLOS Biology, Veller and Coop [9] perform a comprehensive evaluation of how different phenomena can lead to bias in GWAS and FGWAS, finding that FGWAS is free from all environmental confounding and almost all genetic confounding.Typically, GWASs analyze one focal variant at a time, so GWAS associations include the direct genetic effects (DGEs, causal effects of alleles in an individual on that individual) of the focal variant and variants that are correlated with the focal variant (i.e., in linkage disequilibrium or LD).Variants that are physically close (on the same chromosome) to the focal variant tend to be in strong LD as they tend to be inherited from the same parental haplotype without recombination.This makes it hard to pinpoint the causal variant, leading to the problem of fine-mapping.Since this occurs even under random-mating, it's not typically thought of as