Tree-based QTL mapping with expected local genetic relatedness matrices.

Area of Science:

• Population genetics
• Statistical genetics
• Genomics

Background:

• Genome-wide Association Studies (GWAS) independently test genetic variants, overlooking their correlated evolutionary history.
• The Ancestral Recombination Graph (ARG) models this shared evolutionary history using local coalescent trees.
• Estimating ARGs from large-scale samples is now computationally feasible.

Approach:

• We propose a quantitative trait locus (QTL) mapping framework using the conditional expectation of a local genetic relatedness matrix given the ARG (local eGRM).
• This ARG-based approach models the shared evolutionary history of genetic variants.
• The method is framed within existing variance-components approaches.

Key Points:

• Simulations demonstrate the method's advantage in detecting QTLs with allelic heterogeneity.
• The ARG-based approach facilitates QTL detection in understudied populations.
• We identified a large-effect BMI locus (CREBRF) in Native Hawaiians, previously undetectable by GWAS.

Conclusions:

• The local eGRM framework effectively leverages estimated ARGs for QTL mapping.
• This approach improves the power to detect genetic associations in complex traits.
• Our findings highlight the utility of ARG-based methods in population and statistical genetics.