Many studies start (left) with large human cohorts and a wealth of data. These data can be used to formulate a working hypothesis (e.g., the microbiome can cause some of the diabetes phenotype). Then, the researcher moves to a host-focused proof of concept (top) to validate the hypothesis (e.g., FMT from humans with diabetes to GF mice phenocopies poor glucose tolerance). Then, the researcher begins to understand the mechanisms underlying this causative effect of the microbiome (right) using targeted models (e.g., monocolonization with a suspected key microbe, deep phenotyping of cellular responses in highly controlled experiments). With a focal mechanism, an intervention can be constructed and tested preclinically (bottom) to modify the affected mechanism identified in previous phases (e.g., a pre-, pro-, or post-biotic target that can restore missing metabolites needed in the, e.g., insulin-transduction pathway). If safety and efficacy are observed, the next step is a clinical trial. If the clinical trial is less promising than the preclinical experiments, iterative testing may be needed to determine dose, possibilities of co-administration (e.g., synbiotics), and other nuances.
