3/9 – Christopher Henry, Argonne National Laboratory

Speaker: Christopher Henry

Computational Biologist, Argonne National Laboratory

Title: Building KBase to Enable Mechanistic Understanding from a Molecule to an Ecosystem

Abstract:

Biology today brings enormous challenges. Only 10% of metabolites observed in the environment with mass spec can be identified; numerous genes have persistently unannotated function across all domains of life; we lack an understanding of how microbiome systems assemble, adapt, and function; and for all of these reasons, we continue to be unable to account for contribution of microbial and other forms of life when we model ecosystem dynamics. Yet, today great opportunities abound. Mass spec instruments are getting better; genomics and metagenomics data is exploding; machine learning is making unintuitive leaps in developing inferences from data; and models are getting better. To address these challenges and leverage these opportunities, we need software tools that integrate data across scales, embedding data into mechanistic models capable of rigorously testing, validating, and ultimately expanding knowledge. The DOE Knowledgebase (KBase) is being developed to achieve this goal. As demonstrations, I will discuss some science we have performed using the KBase platform, including: (i) mechanistically explaining natural microbiome progression on built surfaces; (ii) reconciling genome annotations and pathways to exometabolite data to discover conserved ecosystem services performed by Pseudomonas; (iii) applying cheminformatics to predict novel pathways and propose putative structures to unidentified peaks in FTICR metabolomics data; and (iv) large-scale construction and comparison of metabolic models for MAGs and entire microbial communities based on environmental metagenome data. Across all of these stories we will explore how mechanistic modeling is a valuable tool for moving beyond correlation to experimentally testable hypotheses.