Talk by Jennifer Listgarten: How to condition your favorite sequence model, for protein engineering and beyond

Abstract: I'll go over two of our recent papers that are related to each other. In the first, we show how to correctly do guidance for diffusion and flow matching on discrete state spaces, such as protein sequences. In the other, we extend this guidance to Masked Language Models and Any Order Autoregressive models, enabling us to apply guidance to increase the activity of a base editor.

Bio: Jennifer Listgarten is a Professor in the Department of Electrical Engineering and Computer Science, the Center for Computational Biology, and the Bioengineering program at the University of California, Berkeley where she holds the Jeffrey Huber and Angel Vossough Chancellor’s Chair in Computational Biomedicine. She is also a member of the steering committee for the Berkeley AI Research (BAIR) Lab. From 2007 to 2017 she was at Microsoft Research, through Cambridge, MA (2014-2017), Los Angeles (2008-2014), and Redmond, WA (2007-2008). She completed her Ph.D. in the machine learning group in the Department of Computer Science at the University of Toronto, located in her home town. She has two  undergraduate degrees, one in Physics and one in Computer Science, from Queen's University in Canada. Jennifer's research interests are broadly at the intersection of AI/machine learning, applied statistics, molecular biology and science. Her current research is primarily in understanding how machine learning can be used to advance protein engineering.

Note: Jennifer also gave a seminar earlier this week in the Center for Protein Design. There will be some overlap to that talk, although she will focus more on the technical aspects here.