Genomic variance estimates: With or without disequilibrium covariances?

Area of Science:

• Plant genetics
• Quantitative genetics
• Genomic analysis

Background:

• Whole-genome regression methods estimate heritability by relating phenotype to marker genotypes.
• Debate exists on incorporating linkage disequilibrium (LD) into genomic variance estimation.
• LD can significantly influence estimates of heritability and genetic variance.

Purpose of the Study:

• To investigate two genomic variance estimation methods with differing LD accounting capabilities.
• To analyze the contribution of quantitative trait loci (QTL) covariances to genomic variance.
• To assess the impact of population structure on LD and variance estimates.

Main Methods:

• Analysis of flowering time in 1,057 sequenced Arabidopsis lines.
• Comparison of a classical genomic variance estimate (ignoring LD) with an LD-explicit method.
• Utilizing principal component analysis (PCA) to account for population structure.

Main Results:

• Genomic variance estimates that ignore LD covariances between QTL underestimated the true genomic variance.
• The LD-explicit method yielded estimates that, when combined with error variance, matched observed phenotypic variance.
• Strong LD and significant covariances were detected between Arabidopsis chromosomes, influenced by population structure.

Conclusions:

• Accurate genomic heritability estimation requires accounting for LD, particularly covariances between QTL.
• Population structure can induce substantial LD, affecting variance estimates.
• LD-explicit methods are essential for robust genomic variance estimation in structured populations.