Leveraging ancestral recombination graphs for scalable mixed-model analysis of complex traits

Area of Science:

• Genomics
• Statistical Genetics
• Computational Biology

Background:

• Genome-wide ancestral recombination graphs (ARGs) model genealogical relatedness.
• ARGs can capture effects of unobserved genetic variants.
• Previous methods were computationally prohibitive for large-scale analyses.

Purpose of the Study:

• To develop computationally efficient methods for using ARGs in large-scale genetic analyses.
• To enable the estimation of heritability and perform association testing using ARGs.
• To improve the discovery of gene-trait associations.

Main Methods:

• Developed sublinear time genotype-matrix multiplication leveraging ARGs.
• Implemented scalable randomized algorithms for mixed-model analyses.
• Introduced ARG-RHE, a randomized Haseman-Elston approach for heritability and association testing.

Main Results:

• Demonstrated computational efficiency and statistical power through simulations.
• Applied ARG-RHE to UK Biobank data (337,464 participants, 52 blood traits).
• Identified 8% more gene-trait associations compared to imputation alone.

Conclusions:

• Genome-wide genealogies (ARGs) can effectively complement genotype imputation.
• ARG-RHE offers a scalable and powerful approach for complex trait genetic analyses.
• This method enhances the discovery of genetic associations for quantitative traits.