Inferring interactions in microbe-phage communities

Microbes are found in high abundances in the environment and in human-associated microbiomes, often exceeding one million cells per milliliter. Viruses of microbes, or phages, are found in even higher abundances. Phages are estimated to turn over 10 to 40 percent of microbes daily, contributing to microbial mortality and redirecting nutrient flow between trophic levels. Understanding the ecological dynamics of microbial communities requires knowledge of phage “host-range” i.e. which microbes a phage can infect. However elucidating phage host-range in situ remains a difficult and open question. Here, we discuss current approaches to the host-range inference problem and benchmark their performance using in silico microbe-phage communities. Contrary to widespread use, correlation and correlation-based approaches on abundance time-series do not recapitulate phage host-range (Coenen and Weitz, 2017). Model-based approaches, which compute regressions on discretized mechanistic models, can accurately recover host-range and are robust to variation in network structure and life history traits (Jover et al, 2016). Finally, we extend the model-based approach to simultaneously infer heterogeneous microbe-microbe competition in the presence of phages.