Computational resources for drug discovery are being heralded as a new, rapid, and efficient approach for drug discovery. Structure based drug design (SBBD) along with virtual high throughput screening (vHTS) have resulted in several clinical drug candidates. However, the overall complexity of drug discovery and lack of innovative ideas has led to the stagnation of success. Such limitations have been experienced in de novo protein structure prediction and an avant-garde approach has been tested to overcome these limitations: crowd-source of protein folding through an online multiplayer game. Foldit leverages human problem solving skills to decipher protein folding problems set up as puzzles. Unlike the use of computational approaches alone, collaboration between players relies on intuition, creativity, and motivation of players to solve complex scientific problems. Players not only explore the conformational space of proteins but also enumerate possible search strategy algorithms, even those not implemented in structure prediction suites. Using new novel algorithms, Foldit players have successfully determined the structure of a retroviral protease. We believe that leveraging human intuition through the creation of a scientific game for drug discovery can not only provide a new avenue to discover therapeutic drugs for protein targets but can also contribute to the development of new sophisticated algorithms to use in computer-aided drug design.
The objective is to expand Foldit with the ability to dock and design drug-like compounds through a fragment-based approach. Fragments identified through high throughput screening HTS and vHTS will be given to players. The discrete set of fragments will circumvent combinatorial explosion of possible small-molecules built. Scoring methods that incorporate free energy of binding and physicochemical properties of the ligand will help guide the players’ designs. The accuracy of the players’ models will be assessed through a preliminary proof of concept for a previously designed fragment-based inhibitor of VEGF receptor 2. Once utility of Foldit for fragment-based drug design is proven, candidate receptors for new puzzles will be identified in collaboration with Eli Lilly. Players will then be allowed to create drugs for the candidate receptors. Player created small-molecules will be assessed for drug-like properties and ease of synthesis. Drugs that pass these metrics will be synthesized and screened for activity through nuclear magnetic resonance (NMR) experiments, which allows for rapid detection of binding affinities for weak binding molecules. Players who design successful bound molecules will be interviewed to better understand the strategies which they used to design drugs. These insights will then be implemented in algorithms in Rosetta and Foldit to further improve computational docking and drug design of small molecules.
Current Project Members: Oanh Vu
Alumni Project Members: Steven Combs