Project Summary The human gut microbiome is associated with a range of diseases, and may positively or negatively affect the success of therapies. However, the causal directions between the human gut microbiome and host health are seldomly clear due to a lack of feasible controlled experiments in humans. High resolution, high frequency temporal data of paired microbiome and physiological measurement, and rich metadata of potential confounders, allow the application of causal inference frameworks. Such data can be mined for potential microbiome drivers of host health, especially if both the microbiome and host physiology are perturbed during the time courses. We have recently published a vast longitudinal microbiome set from cancer patients undergoing severe perturbation of their immune system. Concurrently, these patients experience dramatic shifts in their gut ecosystem. We here propose to unlock this data and build a discovery platform for microbiome causality, towards rational microbiome engineering. We will develop a new machine-learning technique that enables rapid exploration of our data through effective visualization and a web-based interface. We will develop a new method to identify gut microbial competitor species of common pathogens, which are systematically missed by existing approaches but are representing the most promising targets for microbiome engineering. Finally, we will elucidate the bidirectional interplay between human- targeted medications and the gut microbiome. For this, we will leverage our large data set of paired microbiome and host immune cell trajectories. This will validate recent in vitro results indicating that human-targeted medications may influence gut ecology using in situ data, and it will identify potential gut microbiome modulation of pharmacokinetics.