I will outline our advances in learning how to explore molecular space with specific property optimizations or library design challenges in mind. Our very early ideas aimed at optimizing properties by varying molecular Hamiltonians. More recent ideas have shown that optimizing the external potential provides a natural entrée to molecular design; the linear combination of atomic potentials and gradient-directed Monte Carlo approaches resulted from this line of thinking. Our ongoing research aims to explore truly vast molecular spaces for diversity-oriented molecular library design. We are in the process of designing these libraries targeted toward drug discovery and development. The project, supported by NIH, is being conducted in close collaboration with Julia Contreras-Garcia, Xiangqian Hu, Shahar Keinan, Chetan Raj Rupakheti, Aaron Virshup, Peter Wipf, and Weitao Yang. In my talk, I will try to minimize the overlap with seminars from collaborating speakers.
Back to Workshop IV: Physical Frameworks for Sampling Chemical Compound Space