Population Genomic Inference with the Ancestral Recombination Graph (ARG) : Recombination Landscape and Ancestral State Reconstruction

Understanding genetic variation within and among populations is a fundamental goal of population genetics, which aims to reconstruct evolutionary history shaped by mutation, selection, genetic drift and recombination. The ancestral recombination graph (ARG) connects the DNA sequences in a sample through historical coalescence and recombination events and contains the complex genealogical relationships between sampled individuals. By capturing these historical relationships, the ARG provides a robust framework for obtaining rich information to answer key population genetics questions, making it an increasingly influential tool in the field.

In this thesis, we develop methods that utilize ARG for two applications: to (1) infer recombination map, (2) infer ancestral states, and estimate the unfolded site frequency spectrum (uSFS), demonstrating the applications of ARG in population genetic analysis.

To infer the recombination map, we introduce a new framework tailored for small sample sizes, utilizing transitions of genealogies along the genome. Our approach is faster than the existing iSMC method for single diploid genomes but comes at some cost of accuracy. 

We develop the first outgroup-free method for ancestral state inference and uSFS estimation. We theoretically characterize when local genealogies contain—or fail to contain—ancestral state information. We evaluate how accurately the ancestral allele and uSFS are inferred when the ARGs are known and when different methods reconstruct ARGs. We apply our new method to simulated data under various scenarios and a human dataset to compare it to the outgroup-based method. Our method provides a powerful alternative or complementary method for polarization.

Overall, this thesis explores the challenges and potential of using genome-wide genealogy information for population genetics analyses and contributes to advancing ARG-based inference methods.

Rechte

Nutzung und Vervielfältigung:


CC BY 4.0

Bitte beachten Sie, dass einzelne Bestandteile der Publikation anderweitigen Lizenz- bzw. urheberrechtlichen Bedingungen unterliegen können.

Zitieren

Zitierform:
Zitierform konnte nicht geladen werden.