Encode AI for Science Fellow (Pillar VC / Imperial College London) Protein Design, PhD in machine learning for structural biology at UCL
London
Joined March 2009
- Tweets369
- Following1.9K
- Followers1.5K
- Likes2.9K
Sorry everyone, that was me trying to align two PETase sequences.
Breaking News: Anthropic said it halted potential plots by scientists who used its A.I. models to do research that could have helped develop biological weapons. nyti.ms/3V3epAE
Great experiment and a great dataset: it shows that Claude can competently use the open-source tools that practitioners would use in a binder design campaign. but most of the magic is in the tools themselves and telling it which tools to use and how (the prompt)
Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is between 10% and 15%.
Between 22% and 35% of Claude's designs bound successfully, depending on the setup. Some of its strongest designs bound several times more tightly than the best published de novo binder.
if I knew I was competing on behalf of Team Human against the AIs, I would have increased our compute budget by 50X to match the amount used by Claude. and optimised more for hits rather than aiming for an informative benchmark set. Blog post coming soon on those 15PGDH results
Jude Wells retweeted
Great experiment and a great dataset: it shows that Claude can competently use the open-source tools that practitioners would use in a binder design campaign. but most of the magic is in the tools themselves and telling it which tools to use and how (the prompt)
Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is between 10% and 15%.
Between 22% and 35% of Claude's designs bound successfully, depending on the setup. Some of its strongest designs bound several times more tightly than the best published de novo binder.
Interesting that the ESMfold2 paper shows that it still needs MSAs to match the performance of the models that came out more than a year ago (AF3 & Protenix). Still great to see progress towards MSA-free structure prediction. [results on FoldBench monomer prediction]
Jude Wells retweeted
3/11
Not because it was always the goal, but because the incentives push them there.
And those incentives are hard to ignore.
Therapeutics dominate the biggest exits.
From 2009–2023, Pear VC analyzed the top 50 life sciences startups with exits >$2.7B:
92% therapeutics (46/50)
6% diagnostics (3/50)
2% tools companies (1/50)
We've been working on protein design with hallucination, and it turns out there's some nice optimisation theory behind this approach, beautifully explained by my colleague Magnus here:
magnusross.github.io/posts/r…
Jude Wells retweeted
next insilico is next wednesday! 🧠💊💾🍳
speakers include: Ruxandra Teslo (Renaissance Philanthropy), Raina Thomas (Google DeepMind), Thomas Löhr (Bind Research), and Dominykas Lukauskis (Ternary Therapeutics)!!
hosted at Fifty Years HQ 🇬🇧
luma.com/2daur0ny
