Data-driven prediction of colonization outcomes for complex microbial communities

Complex microbial interactions can lead to different colonization outcomes of exogenous species, be they of pathogenic or nonpathogenic in nature. Predicting the colonization outcome of exogenous species from the baseline compositions of complex communities remains a fundamental challenge in microbial ecology, largely due to our limited knowledge of the diverse physical, biochemical and ecological processes governing the microbial dynamics. Here, we proposed a purely data-driven approach that is independent of any dynamics model to predict colonization outcomes of exogenous species for complex microbial communities. We systematically validated this approach using synthetic data, finding that it can predict not only the binary outcome but also the final abundance of the invading species. Then we conducted in vitro colonization experiments of two commensal human gut bacteria species Enterococcus faecium and Akkermansia muciniphila in hundreds of human stool-derived in vitro microbial communities, confirming that our approach can successfully predict the colonization outcomes. The presented results suggest that the data-driven approach could be a powerful tool to inform the management of complex microbial communities.