NVIDIA AI Podcast From AlphaFold to MMseqs2-GPU: How AI is Accelerating Protein Science - Ep. 273
74 snips
Sep 10, 2025 Chris Dallago, Research Lead at NVIDIA and Visiting Professor at Duke University, teams up with Martin Steinegger, co-author of the Nobel Prize-winning AlphaFold paper and Associate Professor at Seoul National University. They dive into the revolutionary impact of AI on protein science, discussing how GPU acceleration transforms protein structure prediction and drug discovery. From innovative models like AlphaFold to the game-changing MMseqs2-GPU, they emphasize the urgency of efficient data processing and the value of collaborative research in biological advancements.
AI Snips
Chapters
Transcript
Episode notes
Structure Determines Function
- Protein 3D structure determines biological function and drives cellular machinery.
- Knowing structure is essential for drug discovery and modifying proteins for useful tasks.
AlphaFold Changed The Field
- AlphaFold and related models transformed biology by enabling reliable computational folding.
- Labs and industry now routinely use predicted structures to accelerate discovery and research.
Homology Gives Folding Constraints
- Homology retrieval gathers evolutionary-related sequences to build MSAs that constrain folding.
- Those multiple sequence alignments reveal conserved and contact patterns that make structure prediction easier.