16S data from FMT trial for the treatment of recurrent Clostridium difficile infection

Fecal microbiota transplantation (FMT) is a treatment for microbiome-associated diseases in which gut microbiota are transferred from a healthy donor to a patient. Although the success of FMT requires donor bacteria to engraft in the patient's gut, the forces governing bacterial engraftment in humans are unknown. Here, we use a vast, ongoing clinical experiment - the treatment of recurrent Clostridium difficile infection with FMT - to uncover the rules of engraftment in humans. First, we built a machine learning model that accurately predicts which bacterial species will engraft in a given host. We then developed a maximum-likelihood strain inference method, Strain Finder, allowing us to infer the genotypes of donor strains and to track them through patients' guts over time. Surprisingly, engraftment could be predicted largely from the abundance and phylogeny of bacteria in the donor and the pre-FMT patient. We also found that donor strains within a species engraft in an all-or-nothing manner and that previously undetected strains frequently colonize the patient after FMT. We validated these findings in another disease context, metabolic syndrome, suggesting that the same principles of engraftment extend to other indications. These findings may guide the design of bacterial therapeutics that target diseases ranging from ulcerative colitis to cancer.