Uploaded May 2026 | Updated September 2026, 2 weeks ago
Eric and Wendy Schmidt Center Symposium: Biomedical Science and AI
April 28 - 29, 2026
Day 1, Invited talk: The role of intrinsically disordered regions (IDRs) in directing transcription factor binding vast genomes
Jennifer Listgarten, UC Berkeley
Abstract:
First, I'll discuss how to take an already-trained sequence generative model—such as ESM3 or ProteinMPNN (or any masked language model, autoregressive model, any-order autoregressive model, diffusion model, or flow matching model)—and tailor it on-the-fly to your specific use case without retraining the original model. Specifically, I'll show how to statistically condition a pre-trained model on properties it wasn't trained for by "guiding" the sampling process with a predictive model. This guidance can come from newly collected experimental data, a biophysical model, or any other predictive framework. We used this method to re-engineer a base editor, finding that a single round of guidance achieves a higher editing efficiency than was previously achieved using seven rounds of directed evolution. Second, I'll discuss our recent work probing a model affibody system to better understand how two proteins bind each other, examining epistasis, the topology of the fitness landscape, and its coupling with structural geometry. We accomplish this using a combinatorially complete fitness landscape measured with a protein library-on-library system over six rounds of selection, analyzed with statistical machine learning methods.
Speaker 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 and an ISCB Fellow. From 2007 to 2017 she was at Microsoft Research. 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. Her undergraduate degrees were in Physics and Computer Science, from Queen's University in Canada. Jennifer's research interests are broadly at the intersection of machine learning/AI, applied statistics, and biology. Her current research primarily focuses on understanding how AI can be used to advance protein engineering.
For more information, visit broad.io/ewsc-symposium-2026
Copyright Broad Institute, 2026. All rights reserved.
Eric and Wendy Schmidt Center Symposium: Biomedical Science and AI
April 28 - 29, 2026
Day 1, Invited talk: The role of intrinsically disordered regions (IDRs) in directing transcription factor binding vast genomes
Jennifer Listgarten, UC Berkeley
Abstract:
First, I'll discuss how to take an already-trained sequence generative model—such as ESM3 or ProteinMPNN (or any masked language model, autoregressive model, any-order autoregressive model, diffusion model, or flow matching model)—and tailor it on-the-fly to your specific use case without retraining the original model. Specifically, I'll show how to statistically condition a pre-trained model on properties it wasn't trained for by "guiding" the sampling process with a predictive model. This guidance can come from newly collected experimental data, a biophysical model, or any other predictive framework. We used this method to re-engineer a base editor, finding that a single round of guidance achieves a higher editing efficiency than was previously achieved using seven rounds of directed evolution. Second, I'll discuss our recent work probing a model affibody system to better understand how two proteins bind each other, examining epistasis, the topology of the fitness landscape, and its coupling with structural geometry. We accomplish this using a combinatorially complete fitness landscape measured with a protein library-on-library system over six rounds of selection, analyzed with statistical machine learning methods.
Speaker 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 and an ISCB Fellow. From 2007 to 2017 she was at Microsoft Research. 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. Her undergraduate degrees were in Physics and Computer Science, from Queen's University in Canada. Jennifer's research interests are broadly at the intersection of machine learning/AI, applied statistics, and biology. Her current research primarily focuses on understanding how AI can be used to advance protein engineering.
For more information, visit broad.io/ewsc-symposium-2026
Copyright Broad Institute, 2026. All rights reserved.
