SBIR Phase I: Leveraging machine learning to enable generalized phage therapy for pulmonary infections

The broader impact /commercial potential of this Small Business Innovation Research (SBIR) Phase I project is to develop a new therapy for bacterial infections, especially those resistant to current antibiotics, which have generated antibiotic-resistant “super-bug” bacterial infections that cannot be treated easily. Bacteriophages (‘phages’) are viruses that only infect specific bacteria and cannot infect humans.

Phages kill harmful bacteria, but they currently do not work well as general solutions that can be prescribed broadly because each phage only kills a subset of bacteria; therefore a unique phage may be required for different people with the same infection. This project develops new technology to understand how phages target bacteria. It uses machine learning to determine the parts of each phage responsible for killing specific bacteria, in order to make phages for broad use in treating infections.

This innovation is a key competitive advantage, and helps both national health and defense by creating new treatments for antibiotic-resistant infections, which cost >$64 billion annually and may become the next major pandemic.

This Small Business Innovation Research (SBIR) Phase I project will develop machine learning algorithms that identify genetic determinants of host range in phages in order to engineer phage to have expanded host range. The widespread evolution of multidrug-resistant infections is a major threat to global health, and traditional antibiotics have significant adverse effects on patients and their microbiomes.

Phages can solve this global health challenge, but the inability to expand and tune phage host-range to create a generalizable therapeutic remains a key barrier to commercial success. This project will leverage machine learning and proprietary high throughput phage characterization methods to generate maps of phage-host interactions to identify genes that determine phage host range, and use novel engineering techniques to validate these genetic determinants of host range.

The expected outputs are twofold: 1) a machine learning model for predicting variants, genes, or genomic regions that determine phage host range and 2) an engineered phage with expanded host range. This work will further scientific understanding of phage biology and phage-host interactions, while also providing a platform to develop phages with tunable host range for therapeutic, agricultural, and environmental applications.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.