Unified Computation Tools for Natural Products research

Project Summary/Abstract Accurate, reliable and efficiently obtained structure determinations and annotations are of increasing importance in the current era of integrated omics research. Natural products continue to provide inspirational new structures for drug discovery programs, and genomic studies of microorganisms, including our work with marine cyanobacteria, have revealed that there are many times more secondary metabolites genetically encoded than are typically expressed in cultures. Building on this knowledge, new methods in elicitation and heterologous expression are successfully accessing this previously hidden genetic capacity for natural products biosynthesis. Metabolomic profiling interfaced with improved molecule annotations has richly informed on regulatory mechanisms of expression and the role of microbial interactions, especially including those involved in symbiotic interactions. This continuing NIH supported program has developed several of the most widely used annotation and analysis tools for MS and NMR natural product datasets, including GNPS, MASST, MassQL, SMART 2.1 NMR, DeepSAT, NP Classifier, NPOmix, SMART-Miner and PECAN, among others, and has led to 41 published papers in the previous 4-year grant cycle. These publicly available tools have changed the way that diverse scientists observe, organize, and make sense of complex metabolic profiling datasets from LC/MS and NMR analyses. For example, over the last 12 months, the GNPS ecosystem of tools has been used by >50K users from 156 countries and over the lifetime of the GNPS ecosystem, the community has analyzed over 25M LC/MS samples. SMART 2.1 NMR and DeepSAT have been used by 2.3K users in 66 countries. Collectively, these tools have democratized access to cutting edge computational approaches and are making an impactful difference in the way that researchers probe small molecule-based interactions of diverse life forms. In this next proposed grant period, we focus on machine learning and AI techniques to enhance the computational approaches we have pioneered and expand the breadth of problems within small molecule analysis. Cottrell and Gerwick will continue to develop annotation tools for even more precise predictions of structures and biological properties from NMR data, including improved mixture analysis. New member Wang will innovate on the integration of MS and NMR-based analyses to disambiguate closely related substances. Also, Wang will continue to leverage extensive computational infrastructure to ensure public and online access to the tools developed in this NIH program, democratizing their use and enhancing their impact. These overarching goals lead to five specific aims that will provide the community of diverse chemists and biologists a set of easily accessible tools for the improved identification of organic molecules and their properties by a variety of NMR and MS based methodologies. These tools will be based on the most current methods available for machine learning and artificial intelligence, thus successfully integrating computer science with analytical chemistry.