Moonshine.jl: a Julia package for genome-scale model-based ancestral recombination graph inference

Area of Science:

• Statistical genetics
• Population genetics
• Computational biology

Background:

• The ancestral recombination graph (ARG) is a key model for understanding population ancestries.
• Inferring ARGs from haplotype data has lagged behind ARG simulation methods.
• Previous methods struggled with large, real-world genomic datasets.

Purpose of the Study:

• Introduce Moonshine.jl, a Julia package for efficient ARG inference.
• Enable model-based inference for large-scale haplotype data.
• Improve the scalability and accessibility of ARG inference.

Main Methods:

• Developed Moonshine.jl, a Julia package utilizing threading-based algorithms.
• Implemented restrictions on probability distributions for sample consistency.
• Optimized for inferring ARGs on samples of thousands of human haplotypes.

Main Results:

• Moonshine.jl can infer ARGs for up to 10,000 densely haplotyped human chromosomes (hundreds of megabases) in under a day.
• Achieved efficient, model-based inference for large datasets.
• Demonstrated scalability for compute cluster deployment.

Conclusions:

• Moonshine.jl significantly advances the capability for large-scale, model-based ARG inference.
• The package offers efficiency, ease of use, and integration into biostatistical workflows.
• Facilitates deeper insights into population genetics and human ancestry.