Abstract: Practical Quantum Computing for Peptide Drug Discovery Rational drug design methods require clever ways of searching the enormous space of possible molecules for the small handful that might have desired physical, chemical, or pharmacological properties needed to treat some disease. Alongside modern machine learning methods such as deep neural networks, quantum computers represent one of the most exciting new technologies poised to revolutionize computational drug design. Quantum mechanical systems can exist in superpositions of enormous numbers of states, each of which can encode a different piece of information -- a possible solution to a hard search problem, for example. A computer built to harness quantum mechanical principles can operate on all of these states simultaneously, giving rise to a sort of massive parallelism that could potentially allow currently intractable search problems to be solved. In this talk, I will discuss my current work to apply current and near-future generation quantum annealers to hard problems in biomolecular modelling, with a particular focus on drug design. I will show examples of molecules designed on the quantum annealer for which we now have experimental validation, and cover current challenges that we hope to overcome to improve our success rate. I will also cover current attempts to adapt these methods for current and near-future gate-based quantum computers. Finally, I will discuss prospects for using quantum computers in conjunction with machine learning to enhance the success rate of rational drug design.

Research Interests: Computational methods for designing folding heteropolymers, using modern machine learning approaches on classical computers, plus new quantum algorithms on current-generation quantum computers.

Bio: Vikram Khipple Mulligan completed his Ph.D. in the Dept. of Biochemistry at the University of Toronto, working to elucidate the molecular mechanisms of protein misfolding in late-onset neurodegenerative diseases like amyotrophic lateral sclerosis using time-resolved spectroscopic techniques. He then worked as a postdoc in the Dept. of Biochemistry at the University of Washington, generalizing the Rosetta software suite, which had to that point been used primarily for predicting structures of and designing proteins, to allow structure prediction and design of synthetic heteropolymers built from artificial chemical building blocks. As a Research Scientist in the Center for Computational Biology at the Flatiron Institute, Dr. Mulligan is continuing to develop advanced computational methods for rational design of peptide therapeutics, combining physical simulation with machine learning methods and with quantum computing to make hard problems in peptide design tractible.