AI-designed protein starting points outperform natural enzymes in directed evolution
bravo_abad · x · 2026-07-28
What the study shows
Researchers from the Broad Institute and Harvard used ProteinMPNN to redesign starting sequences for botulinum neurotoxin proteases, while constraining catalytic and substrate-binding residues.
Key results
- The redesigned proteins kept full catalytic efficiency.
- They also expressed better and melted at temperatures up to 9°C higher.
- In side-by-side phage-assisted continuous evolution experiments, the AI-redesigned starting points consistently outperformed the natural enzyme under the same selection conditions and codons.
- Across three substrates of increasing difficulty, the redesigned variants reached higher final activity every time.
Why it matters
The paper’s core claim is not just that AI can design protein function, but that it can produce better starting points for directed evolution, improving the outcome of the evolutionary search before it even begins.
Related event: Nature Study: AI-Redesigned Proteins Boost Enzyme Evolution(2 posts)→
More from Research
- Xiaohongshu FireRedTTS3: Unified Speech Generation and Editing Model — jiqizhixin · 2026-08-26
- Academics debate ethics of AI-assisted paper authorship — RichmanRonald · 2026-08-26
- Anima Anandkumar's Startup Covered by Reuters: Training 4D Physics-Simulating AI Models — iScienceLuvr · 2026-08-26
- Anima Anandkumar's new startup trains trillion-context models that understand physics in 4D — xiaohu · 2026-08-26
- CommerceAgentBench Grades Agents on Operational Traces, Not Self-Reports — and Hides BEC Fraud in the Inbox — alifcoder · 2026-08-26
- Discrete diffusion to replace speculative decoding: Research preview — LucaAmb · 2026-08-26