An Open Web Service for Fragment-based Design of Small Molecule Inhibitors

Period of Performance: 06/01/2016 - 05/31/2017


Phase 1 SBIR

Recipient Firm

Conifer Point Pharmaceuticals, LLC
Principal Investigator


? DESCRIPTION (provided by applicant): Maps of where small chemical fragments bind to proteins are proving to be an important source of information for life sciences research, most particularly in drug discovery. For example, such maps are used to: (1) discover binding hot spots by fragment clustering; (2) identify how fragment binding patterns differ among protein structure variations, critical for achieving selectivity, mutation avoidance, or multi-targeting; ad (3) find starting points for difficult targets. The best computationally-generated fragment maps are produced by Grand Canonical Monte Carlo fragment-protein simulations using simulated annealing of chemical potential (GC/MC-SACP). These maps, the subject of numerous publications, allow predictive free energy ranking of binding sites, incorporate comprehensive sampling of binding configurations, and describe rigorous Boltzmann distributions. Our primary assertion is that making these fragment (and water) binding maps and tools widely available at low cost, will have a transformational impact on the productivity of researchers and drug designers. Foundations: Historically, these simulations have been considered expensive, but we have developed a highly optimized GC/MC-SACP simulator to dramatically reduce the cost. Over the last 5 years, we have computed an unprecedented ~250,000 GC/MC-SACP fragment and water maps (GMAP's). We have used these maps in identifying and optimizing drug lead compounds in over a dozen projects. This includes several protein-protein interaction inhibitors, targeting PCSK9 (cholesterol-lowering) and RecA (bacterial DNA repair). Goal: We will make GMAP's and fragment-based design tools available to the broadest possible research community at low cost using recently available Web and Cloud computing technologies. GMAP's are currently not available to most researchers. Our tools currently require optimized software on high-end workstations. Innovation: GMAP's consist of 100's GB per protein. We will use matching pursuit algorithms for statistical modeling to derive compact representations of fragment distributions. This is novel and a publishable result. The computational demands of interactive fragment-based design can now be met by employing the WebGL (3D graphics) and WebCL (physics computation) API's in browsers, allowing compute-intensive functions to be run on the user's graphics hardware (GPU). This is the future of computational chemistry tools Aim 1: Use matching pursuit algorithms to achieve 100x reduction in the size of GMAP's, compress our existing maps for several hundred therapeutically-relevant proteins, and demonstrate practical delivery of GMAP's over the Internet. Aim 2: Implement a preliminary Web service for fragment visualization, for searching GMAP's for bondable fragment pairs, and for clustering fragments for hot-spot analysis. Overall Impact: The full program will deliver a tiered Web/Cloud service. Researchers need alternative approaches to drug lead development. Mining GMAP data is a fertile source of new chemistry ideas. When widely adopted, the comprehensive service would provide the platform to significantly advance drug discovery.