Genome-wide association studies of complex traits often are complicated by relatedness among individuals. Ignoring or inappropriately accounting for relatedness often results in inflated type I error rates. Either genotype or pedigree data can be used to estimate relatedness for use in mixed-models when undertaking quantitative trait locus mapping. We performed simulations to investigate methods for controlling type I error and optimizing power considering both full and partial pedigrees and, similarly, both sparse and dense marker coverage; we also examined real data sets. (1) When marker density was low, estimating relatedness by genotype data alone failed to control the type I error rate; (2) this was resolved by combining both genotype and pedigree data. (3) When sufficiently dense marker data were used to estimate relatedness, type I error was well controlled and power increased; however, (4) this was only true when the relatedness was estimated using genotype data that excluded genotypes on the chromosome currently being scanned for a quantitative trait locus.
Keywords: quantitative trait locus (QTL); relatedness; statistical power; type I error rate.