Quantum Computing and AI Models Larger Proteins to Find New Drugs

ProteinQure uses quantum computing, molecular simulations and reinforcement learning to engineer new drugs.

They combine state-of-the-art structure-based algorithms, including molecular dynamics simulations and machine learning. Their design platform explores vast regions of sequence space and predict functional properties of therapeutics to rapidly deliver new insights to their partners in pharmaceutical R&D.

ProteinQure can obtain structures for protein therapeutics and drug targets (up to ~100 amino acids). Our integrative models can use external data (sequence, structure, and functional measurements) to increase the speed and accuracy of their predictions. This approach has shown strong agreement to experimental structures in blind protein folding challenges (CASP).

ProteinQure is applying these methods to develop a novel class of peptide-mimetic polymers with SRI International. These molecules have shown high binding affinity to well-established drug targets involved in cell signaling (cytokines, nucleotide-binding proteins). By modeling the combinatorial space containing non-natural amino acids, we support rational drug design and optimization of these peptides for stability and binding affinity.

They are world-leaders in quantum computing algorithms for structural biology. Their biomolecular models can integrate outputs from quantum computers to accelerate CPU or GPU-based methods routinely used in structure-based drug discovery, such as molecular similarity and conformational search.

They have published three quantum computing algorithms in the areas of protein folding and molecular docking, with results that match the accuracy of classical solvers. Within 2-3 years, their algorithms will scale with hardware improvements to enable the design of larger proteins modalities outside the reach of modern supercomputers. They have partnerships with Microsoft, D-Wave, IBM, Rigetti, Xanadu and Fujitsu.