Project Summary/Abstract Fungal and bacterial pathogens are a major threat to human health. Few therapeutics exist to treat fungal infections while bacteria are becoming increasingly resistant to existing therapeutics. Humans have been using natural products to treat infections for thousands of years, long before the causal agents of infection were understood. Natural products have continued to be used as therapeutics in the modern age of medicine. Rates of rediscovery of known natural products have increased in traditional sources of natural products, such as soil bacteria. Recently, symbiotic Actinobacteria from insect agricultural systems have been recognized as a promising source of bioactive compounds, especially antifungal agents. These bacteria often produce natural products that defend an insect’s fungal crop from pathogenic fungus. The work proposed here will use chemical biology approaches such as phenotypic interaction screens, genomics, and a new bioinformatics approach to systematically search for bioactive natural products produced by Actinobacteria symbionts and other organisms in insect agricultural systems. The first part of this proposal focuses on using existing techniques to identify new bioactive natural products. Phenotypic interaction screens can identify bioactive natural products by determining if a symbiotic bacteria produces a natural product that inhibits the growth of a fungal pathogen and vice-versa. We will then use genomic sequencing, bioinformatics, and heterologous expression to identify and characterize biosynthetic gene clusters (BGCs) that are not expressed in the phenotypic interaction screens. The second part of the proposed work involves the use of a new bioinformatics technique to identify interesting bioactive natural products. Existing bioinformatics techniques identify BGCs and predict the most likely chemical structure of the corresponding natural product. However, they do not conclude anything concerning the functional role that the natural product plays. The technique developed here will use machine learning to predict the function that the natural product fulfills in the ecological context of the organism. This algorithm will facilitate the identification of bioactive natural products with therapeutically relevant functions.