Big Ideas Lab Drug Discovery
Feb 25, 2025
In this engaging discussion, Jim Brase, Felice Lightstone, and Jonathan Allen from Lawrence Livermore National Laboratory share insights on reshaping drug discovery with advanced computing. They explore how high-performance computing and AI are revolutionizing the identification and testing of new treatments, targeting diseases more precisely than ever. Topics include the innovative use of physics-based simulations, the challenges in drug development, and promising advancements like effective drugs for the 'undruggable' RAS protein. Their optimism shines as they announce new molecular designs entering trials.
AI Snips
Chapters
Transcript
Episode notes
Discovery Is The Main Bottleneck
- Drug development follows five stages and typically takes 10–15 years, with discovery as a major bottleneck.
- Livermore focuses on accelerating the first stage—discovery and development—where forming new drugs can take up to five years.
Traditional Discovery Is A Random Search
- Traditional drug discovery uses large chemical libraries and iterative make-test cycles that are slow and resource-intensive.
- That random, experimental search often requires months per cycle and thousands of compounds to find hits.
Physics Plus AI Narrows Candidates
- Livermore runs physics-based atomistic simulations combined with AI to predict molecules that bind protein targets.
- This lets them narrow candidates virtually before synthesizing compounds, saving time and cost.