Speaker: Purushottam Dixit, Assistant Professor, Department of Physics, Genetics Institute, University of Florida
Topic: A novel physics-based dimensionality reduction approach for microbiome data
Abstract: Host associated microbiome(s) and host’s physiology are tightly intertwined. Modulating the microbiome to achieve target host physiological states and vice versa is an exciting avenue for therapeutic interventions. However, while many specific interactions have been established, quantitative approaches that can predict the microbiome composition from the host physiology and vice versa are lacking. To that end, we present two modeling frameworks (1) EMBED: Essential microbiome dynamics. Similar to normal modes in structural biology, EMBED infers ecological normal modes (ECNs), which represent the unique set of orthogonal dynamical trajectories capturing the collective behavior of a community. Importantly, we find that ECNs often reflect specific ecological behaviors, providing natural templates along which the dynamics of individual bacteria may be partitioned. (2) SMbiot: a shared latent model for microbiomes and host physiology. SMbiot is a dimensionality reduction framework that models the covariation among microbial species and physiological variables with a common latent space. Using several examples of health, agricultural, and environmental interests, we show that SMbiot can predictively capture the relationship between the microbiota and host physiology.