Statistical power and heritability in whole-genome association studies for quantitative traits

Natural variation in root exudate composition in the genetically structured Arabidopsis thaliana in the Iberian Peninsula

Plant root exudates are involved in nutrient acquisition, microbial partnerships, and inter-organism signaling. Yet, little is known about the genetic and environmental drivers of root exudate variation at large geographical scales, which may help understand the evolutionary trajectories of plants in heterogeneous environments. We quantified natural variation in the chemical composition of Arabidopsis thaliana root exudates in 105 Iberian accessions. We identified up to 373 putative compounds using ultra-high-performance liquid chromatography coupled with mass spectrometry. We estimated the broad-sense heritability of compounds and conducted a genome-wide association (GWA) study. We associated variation in root exudates to variation in geographic, environmental, life history, and genetic attributes of Iberian accessions. Only 25 of 373 compounds exhibited broad-sense heritability values significantly different from zero. GWA analysis identified polymorphisms associated with 12 root exudate compounds and 26 known genes involved in metabolism, defense, signaling, and nutrient transport. The genetic structure influenced root exudate composition involving terpenoids. We detected five terpenoids related to plant defense significantly varying in mean abundances in two genetic clusters. Our study provides first insights into the extent of root exudate natural variation at a regional scale depicting a diversified evolutionary trajectory among A. thaliana genetic clusters chiefly mediated by terpenoid composition.

The statistical power of genome-wide association studies for threshold traits with different frequencies of causal variants

This study aimed to investigate the effects of incidence rate, heritability, and polygenic variance on the statistical power of genome-wide association studies (GWAS) for threshold traits. Different incidence rates of threshold trait (1, 3, 5, 10, 25, 40, 50, 60, 75 and 90%), heritability (10 and 25%), and polygenic variance ratio (0 and 25%) were simulated separately for common (MAF ≥ 0.05), low-frequency (0.05 > MAF ≥ 0.01), and rare (MAF < 0.01) variants. Association studies were performed by logistic and linear mixed models. The highest statistical powers were observed in common and low-frequency variants with an incidence of 25-50% and 10-40%, respectively, but for rare variants, the highest statistical power was observed at low incidence. For all causal variant frequencies, the estimated heritability decline with an increase in incidence rate. We found high statistical power for traits with high heritability. In contrast, those with a high polygenic variance ratio have lower statistical power to detect common causal variants using a linear mixed model. These results demonstrate that the incidence rate of threshold traits, heritability, and polygenic variance may affect the statistical power of GWAS. Therefore, it is recommended that the effect of incidence rate, heritability, and polygenic variance be considered in designing GWAS for threshold traits.