Sathya Chitturi (PostEra) - DEL Simulator: A Digital Twin for Understanding Machine Learning on DNA-Encoded Libraries

Wednesday 19 Nov 2025, 11:00 → 12:00 America/Los_Angeles

41/1-1158 - Napa (SLAC)

DNA-encoded library (DEL) screening enables the simultaneous testing of billions of compounds against a protein target, accelerating the discovery of new medicines. Recently, DEL data have been used to train machine learning (ML) models that virtually screen databases of purchasable compounds, expanding discovery beyond DNA-compatible chemistry. However, DEL experiments are inherently noisy, and the impact of this noise on ML performance is not well understood. We developed a realistic digital twin that simulates how DELs are designed and tested, allowing researchers to explore how experimental choices and data processing methods influence model accuracy. Beyond improving DEL experiment design, our digital twin establishes a principled paradigm for using simulated data to dissect and strengthen machine learning methods in medicinal chemistry.

Zoom: https://stanford.zoom.us/j/93013081220?pwd=V6NUjf2cfM5GdWAScLmlunohFdaRzi.1