Data from: Dissecting genome-wide association signals for loss-of-function phenotypes in sorghum flavonoid pigmentation traits

Genome-wide association studies (GWAS) are a powerful method to dissect
the genetic basis of traits, though in practice the effects of complex
genetic architecture and population structure remain poorly understood. To
compare mapping strategies we dissect the genetic control of flavonoid
pigmentation traits in the cereal grass sorghum using high-resolution
genotyping-by-sequencing (GBS) SNP markers. Studying the grain tannin
trait, we find that General Linear Models (GLM) are not able to precisely
map tan1-a, a known loss-of-function allele of the Tannin1 gene, with
either a small panel (n = 142) or large association panel (n = 336), and
that indirect associations limit the mapping of the Tannin1 locus to
Mb-resolution. A GLM that accounts for population structure (Q) or
standard Mixed Linear Model (MLM) that accounts for kinship (K) can
identify tan1-a, while compressed MLMs performs worse than the naive GLM.
Interestingly, a simple loss-of-function genome scan, for
genotype-phenotype covariation only in the putative loss-of-function
allele, is able to precisely identify the Tannin1 gene without considering
relatedness. We also find that the tan1-a allele can be mapped with gene
resolution in a biparental recombinant inbred line (RIL) family (n = 263)
using GBS markers, but lower precision in the mapping of vegetative
pigmentation traits suggest that consistent gene-level resolution will
likely require larger families or multiple RILs. These findings highlight
that complex association signals can emerge from even the simplest traits
given epistasis and structured alleles, but that gene-resolution mapping
of these traits is possible with high marker density and appropriate
models.