Automated; model-guided phenotyping to identify metabolite/gene/microbe interactions

Project Summary/AbstractDNA sequencing has spawned the ?microbiome revolution? -- thousands of microbes and a dizzying number ofmicrobial interactions that are associated with human health and disease. Unfortunately, most species in themicrobiome are known only by a (partial) genome. The limited phenotypic data on newly discovered bacteriareveal species that behave unlike any of our model organisms. While genome-scale modeling plays animportant role in understanding the microbiome, the paucity of phenotypic data for most species preventsdetailed simulation of the microbial communities that affect our health.This project will develop an automated system for profiling, synthesizing, and modeling microbial communities.The center of our approach is Deep Phenotyping, an automated robotic platform that performs complex growthexperiments on demand. Data from Deep Phenotyping will be used to train metabolic and statistical models ofthe oral pathogens Streptococcus mutans and Candida albicans to predict conditions that keep both microbesin a nonpathogenic state.