Automating the path from genomes to predicted chemical structures

Trans-AT polyketide synthases generate natural products with a wide range of bioactivities and are encoded in genomes of many chemically unexplored bacteria, suggesting considerable drug discovery potential. The prediction of natural products from trans-AT polyketide synthase genes has proven difficult in comparison to products from other enzymatic assembly lines. This challenge stems from the functional diversity of these extremely complex enzymes, which obscures correlations between the protein architecture and the generated chemical scaffold. In this work, biosynthetic features from these unique and prolific polyketide synthases were implemented in TransATor, a web application that allows for the first time the efficient automated prediction of polyketide structures from genetic sequences. TransATor provides insight into the stepwise construction of a polyketide backbone and enables the quick prioritization and guiding of isolation efforts with a predicted chemical structure at hand. The use of TransATor was incorporated into an isolation pipeline from diverse unusual bacteria that resulted in the discovery of seven new compounds, thus demonstrating its utility for the natural products community.

Link to the publication in external page "Nature Chemical Biology".