SBIR Phase I: Massive, empirical, scalable generation of small molecule and protein interaction data to enable the discovery of medicines at scale with machine learning

The broader impact of this Small Business Innovation Research (SBIR) Phase I project will be to enable the discovery of new medicines at an unprecedented rate, by empowering researchers across the globe to launch discovery programs in weeks at a dramatically reduced entry price. This is in contrast to the status quo of launching a drug discovery program, which costs hundreds of thousands of dollars, requires expensive company resources, and several months of time prior to an initial readout on progress.

The outcome of this new ability may be an increased number of therapeutics that are able to make it through clinical trials and ultimately to patients. The innovation could provide data of interactions between billions of unique small molecules and the entire human proteome. This could enable the creation of sophisticated and predictive machine learning based models to predict novel interactions between small molecules and protein sequences.

The validation and expansion of this approach would allow for continuous improvement of the ability to predict new medicines for emerging diseases.

The proposed project will seek to demonstrate the ability to encapsulate proteins and small molecules in millions of molecular barcoded pico-scale compartments and identify the specific interactions between the proteins and small molecules contained within each of them. The Intellectual Merit of the activity is in the construction of the materials and methods that enable this capability, which reduces the cost of a single protein and small molecule library screen by over 100X.

Current methodologies are limited to screening in much larger formats, leading to higher costs, limited parallelization, and large quantities of material required for the screen to be used. In contrast, the innovation herein could allow for a dramatic reduction in the amount of materials required, enabling the screening of proteins in high parallelization through miniaturization, and consequently reducing cost and time required to generate a comparable sized dataset by orders of magnitude.

The project goals will be to (1) determine with control systems the ability to identify known interactions in the pico-scale format, (2) determine if non-control protein interactions are able to be identified in at 10-plex format, (3) and to determine if protein selective molecules are able to be identified in a single screen.

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.