Uncovering heterogeneous inter-community disease transmission from neutral allele frequency time series

The COVID-19 pandemic has underscored the critical need for accurate epidemic forecasting to predict pathogen spread and evolution, anticipate healthcare challenges, and evaluate intervention strategies. The reliability of these forecasts hinges on detailed knowledge of disease transmission across different population segments, which may be inferred from within-community transmission rates via proxy data, such as contact surveys and mobility data. However, these approaches are indirect, making it difficult to accurately estimate rare transmissions between socially or geographically distant communities. We show that the steep ramp up of genome sequencing surveillance during the pandemic can be leveraged to directly identify transmission patterns between communities. Specifically, our approach uses a hidden Markov model to infer the fraction of infections a community imports from other communities based on how rapidly the allele frequencies in the focal community converge to those in the donor communities. Applying this method to SARS-CoV-2 sequencing data from England and the U.S., we uncover networks of inter-community disease transmission that, while broadly reflecting geographical relationships, also expose epidemiologically significant long-range interactions. We provide evidence that transmission between regions can substantially change between waves of variants of concern, both in magnitude and direction, and analyze how the inferred plasticity and heterogeneity in inter-community transmission impact evolutionary forecasts. Overall, our study high-lights population genomic time series data as a crucial record of epidemiological interactions, which can be deciphered using tree-free inference methods.