Fully AI-designed molecule Dyno-hc9: 7 mutations picked by deep learning, crosses blood-brain barrier
IAmSamFin · x · 2026-11-20
samsinai details Dyno-hc9, designed in 2021 entirely by a deep learning pipeline — humans designed the process, not the sequence. It is 7 all-necessary mutations from wild type, picked with no human supervision from a single round of 200k designs, and uses a novel, previously unknown mechanism to cross the blood-brain barrier into primate brains (per the graphic: 100x NHP brain delivery, 10x liver detargeting, 1x manufacturing). The team stresses it's a high-throughput platform controlling hit rates, not one-shot success — but the find would be impossible without strong generative models, OOD control, and filters. Context: a reply to Andrew Dunn's 3-level framework for what "AI-powered" molecule means, from his Isomorphic Labs raise story.
More from Research
- Fine-tuned Qwen3-VL 8B reads camera shots at 69% accuracy, beating Opus and Gemini — ravisparikh · 2026-11-20
- Victor Taelin's Bend2 talk from SF Lean Meetup explains the rationale behind the language design — pbaylies · 2026-11-20
- Protocol Labs Launches Neuro Atlas: Companies, Milestones and BCI Funding Mapped — melnykowycz · 2026-11-20
- Reddit debate: Should researchers create a Chinese-language AI conference? — Mean_Revolution1490 · 2026-11-20
- Google's Willow wins 'impossible' magic square game 98.3% of the time, on Science cover — sundarpichai · 2026-11-20
- Yilun Du and Dai Bo release EBM tutorial unifying diffusion, RL, and test-time reasoning — burny_tech · 2026-11-20