Tuning the SMC: efficient simulation and the structure of ARGs

Sequentially Markovian Coalescent (SMC) models form a foundation of modern population genetics, facilitating numerous analytical methods. The SMC demonstrates remarkable alignment with a more detailed model called Coalescent with Recombination (CwR) when examining two-locus summaries; however, the influence of SMC on the intricate structural characteristics of Ancestral Recombination Graphs (ARGs) remains less comprehended.

This study introduces a versatile SMC approximation, SMC(k), where a sole parameter k governs the spatial scale allowing common-ancestor events between distinct ancestral segments. SMC(k) encompasses the conventional SMC and SMC as specific instances and approaches CwR as k escalates, thereby striking a harmonious balance between computational efficiency and precision in simulating recombination.

By employing cutting-edge summaries of ARG structure, researchers elucidate that SMC approximations consistently reduce the continuity of ancestral haplotypes throughout the genome, even while maintaining marginal coalescent properties. Furthermore, it is observed that augmenting k progressively restores this extensive ancestral architecture.

To validate their approach, the authors incorporate SMC(k) within msprime—a software package—and demonstrate that, for modest sample sizes, it can expedite whole-chromosome simulations in species characterized by substantial population-wide recombination rates by several orders of magnitude. Lastly, the team employs SMC simulations for chromosome-scale parametric bootstrapping of demographic inference and concludes that the SMC accurately captures uncertainty in SFS-based estimates, albeit with minor modifications as k increases; yet, the nature of these changes is contingent on the specific ancestry-related traits under investigation.