Scalable phylodynamic inference of mutational and epistatic fitness effects with multi-feature birth-death models
Fitness is a complex phenotype determined by many different genomic features, their interactions with each other and the environment. For rapidly evolving microbes like viruses and bacteria, phylodynamic multi-feature birth-death (MFBD) models provide a way to infer the fitness of lineages with different genotypes from reconstructed phylogenies. However, existing phylodynamic methods assume very…
Fitness, a multifaceted trait, emerges from numerous genomic components and their interactions, as well as environmental influences. For swiftly mutating microbes, such as viruses and bacteria, phylogenetic trees derived through multi-feature birth-death (MFBD) models facilitate the estimation of fitness across lineages with distinct genotypes.
Existing phylodynamic methods, however, rely on simplistic models of genotype-to-fitness (G-to-F) mappings, typically disregarding epistatic interactions among genetic features. To address this limitation, researchers have developed a gradient-based Bayesian MFBD framework to estimate the fitness effects of multiple features, encompassing both additive and epistatic G-to-F mappings.
This framework leverages Hamiltonian Monte Carlo for posterior inference and utilizes shrinkage priors to regulate weakly supported fitness effects, circumventing the need for cross-validation.
In simulations, Bayesian inference demonstrated robust performance in recovering additive fitness effects, with the precision and accuracy of the estimates refining consistently with increasing tree size. Estimating epistatic effects, however, poses a greater challenge, as they hinge on the concurrent occurrence of mutations. Shrinkage priors alleviate this issue, enhancing estimation accuracy when interaction effects are mild or sparsely supported.
Simulations employing misspecified additive models revealed that neglecting moderate to strong epistasis can skew individual mutational effect estimations. Applying this framework to a time-calibrated *Escherichia coli* ST131 phylogeny comprising 1763 samples, researchers estimated fitness effects for specific mutations within the quinolone resistance-determining region (QRDR).
These findings underscore the potential of Bayesian phylodynamic inference in unravelling the genomic determinants of fitness while concurrently quantifying uncertainty surrounding mutational and epistatic fitness effects.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
