Invited Talk — John Chodera

My research focuses on bringing physical modeling into the high-throughput biology and genomics era and reimagining the way we develop small molecules for chemical biology and drug discovery. By combining novel algorithmic advances to achieve orders-of-magnitude efficiency gains with powerful but inexpensive GPU hardware, machine learning, and distributed computing technologies, my lab is developing next-generation approaches and open source software for predicting small molecule binding affinities, designing small molecules with desired properties, predicting the drug sensitivity or resistance of clinical mutations, and understanding the detailed structural mechanisms underlying oncogenic mutations. We co-develop the [OpenMM](http://openmm.org) GPU-powered molecular simulation framework, which powers numerous biomolecular modeling and simulation applications using physical modeling and machine learning. As a core member of the Folding@home Consortium, my lab harnesses the largest computing platform in the world—the first to reach an exaFLOP/s—pooling the efforts of a million volunteers around the world to study functional implications of mutations and new opportunities for therapeutic design against cancer targets and global pandemics. I co-founded the http://openforcefield.org, a scientific collaboration funded by the NIH and an industry consortium consisting of dozens of scientists working to develop modern open source infrastructure for building and applying high-quality biomolecular force fields. I am a founding member of the http://postera.ai/covid, a radical open science patent-free drug discovery effort aiming to develop an inexpensive small molecule therapy effective against COVID-19 and future coronavirus. Using automated biophysical measurements, my laboratory collects new experimental data targeted to advance the quantitative accuracy of our methodologies, and gather new insight into drug susceptibility and resistance in kinases and other cancer targets. My work makes extensive use of scalable Bayesian statistical inference, machine learning via probabilistic programming, and information theoretic principles for designing experiments and quantifying error. I am passionate about open science, disseminating scientific best practices, and maximizing research reproducibility.