|
Abstract:
|
Statistics for microbial community sequencing have been well-studied, but substantial gaps remain in combining these measurements with other high-throughput microbiome profiles, such as metatranscriptomics and metabolomics. Such integrated molecular regulatory and biochemical networks are crucial for understanding complex microbial ecologies in situ, especially as they affect human health. I will focus on one of the most flexible model frameworks to date for multi-omic microbiome epidemiology, Gaussian processes (GPs). With appropriate model components, GPs can capture short- and long-term temporal variation (e.g. transcription vs. growth), subject-specific and clustered data (e.g. inter-individual and family microbial structure), multiple epidemiological or environmental covariates (health status, drugs, diet, or nutrient availability), and discover relationships between microbial community membership, gene carriage, transcriptional regulation, and chemical products. I will conclude with a microbiome-tailored GP implementation and its application to studying diet-microbiome-metabolite interactions and their differences between animal models and human populations.
|
