@001TMF profile picture
Tristan Farmer @001TMF

AI for programmable biologics & protein design | Building biological intelligence | Bio-Defense & personalised therapies | GitHub: 001TMF | UoE

Joined February 2026
  • Posts 574
  • Following 105
  • Followers 375
95 Photos and videos
Pinned Post
Tristan Farmer
@001TMF
Jun 15
Anthropic agents in biology highlighted a gap in agents' capabilities with biological databases. We have an internal solution, and though this is a key component of science, we therefore ported a public version to bring determinism to biological agents.
A tale that will be as long as time. Models were limited by AUP.

ALT A tale that will be as long as time. Models were limited by AUP.

Tristan Farmer
@001TMF
Jun 15
x.com/i/article/206658063980…
5
3
17
3,785
Feed back loops! Yes!
Parmita Mishra
@parmita
Jul 3
the future of medicine will not be won by the company with the loudest model it will be won by the company with the best biological feedback loop @precigenetic
0:15
2
296
Excited
Claude
@claudeai
Jul 2
Announcing Built with Claude: Life Sciences, a global virtual hackathon. Join us and @GladstoneInst for a week of researching and building with Claude Science and Claude Code, with a prize pool of $100k in credits.
Event card for Built with Claude: Life Sciences, a global virtual hackathon, July 7–13, with the Gladstone Institutes logo and a screenshot of Claude Science.

ALT Event card for Built with Claude: Life Sciences, a global virtual hackathon, July 7–13, with the Gladstone Institutes logo and a screenshot of Claude Science.

1
11
3,020
What's the point? Prompt: "Find the most conserved epitopes on the Ebola virus glycoprotein"
Claude
@claudeai
Jun 30
Introducing Claude Science, a new app designed with every stage of research in mind. Artifacts traced to their code, environments managed on demand, and 60 optional scientific databases that you can connect. Available now in beta.
1:26
4
2
18
16,513
@adocomplete what can we use it for?
2
488
about 4 startups i can think of are now dead
Claude
@claudeai
Jun 30
Introducing Claude Science, a new app designed with every stage of research in mind. Artifacts traced to their code, environments managed on demand, and 60 optional scientific databases that you can connect. Available now in beta.
1:26
1
2
280
or this might be like claude design and no one actually uses it
1
1
60
What am I allowed to ask Claude in "Claude Science", built for early-stage research
2
558
Will try. But the question is will it work? Every bio question I asked gets restricted.
Claude
@claudeai
Jun 30
Replying to @claudeai
Learn more: claude.com/science
1
4
338
between these two whats meta up to
AI at Meta
@AIatMeta
Jun 29
We’re sharing the next major milestone in our non-invasive brain-to-text decoder research: Brain2Qwerty v2. Building on v1, which was published today in @Nature, Brain2Qwerty v2 is the highest-performing end-to-end pipeline capable of real-time sentence decoding from raw brain signals. It advances beyond character-level performance to decoding words and semantics, enabling accuracy for overall communication. We believe this research has the potential to make a real difference for the millions of people who suffer from brain lesions or disorders that prevent them from communicating. 🧵👇
1
346
Biodefense will be one of the biggest frontiers in the next decade. Right now, defence < offence and that needs to change.
115
Is there any engineers in SF who are cracked at AI x Bio or are they all in Boston? If that is you though send me a message.
193
The only reason to use a PPI model is if your folding model is inaccurate. Boltz-2 is the biggest weakness of Boltzprot-1. Interested in its ability for the top 1%.
Gabriele Corso@GabriCorso
Jun 16
Replying to @GabriCorso
BoltzProt-1: Our most powerful protein design pipeline, powered by BoltzPPI, a new protein-protein affinity model that prioritizes designs most likely to bind. On hard targets, BoltzProt nearly tripled BoltzGen’s VHH hit rate and outperformed other leading proprietary models, while 58% of binders passed a stringent developability panel, matching clinical-stage therapeutic nanobodies.
3
37
6,373
If people want a meaningful PPI model, I'm happy to build one; I just need a partner for a public release. Have some ideas in mind @lucas_nivon?
2
318
Also, if their PPI is just trunk features pooled into an MLP, that gives me even less faith
1
2
326
Tristan Farmer reposted
sk@compchemm
Jun 29
Protenix-v2 weights available now😍 apples to apples comparison from Lucas Nivon post. Protenix ahead of all; AF3, Boltz, OF3-p2 on a blind test set of protein/ligand. So for Co-Folding use minimum Protenix-v2 huggingface.co/TMF001/proten…
5
14
102
7,822
Don’t know how to explain it but Claude is so much smarter when I’m not in the SF time zone. The minute the US wakes up the model feels immediately lobotomised.
126
I don’t think I’m larping hard enough
Lucy Cai
@lucyjcai
Jun 27
Moving to SF means learning about a whole host of tools-- conductor, greptile, gstack, granola, wispr.... It's like getting fluent in a new language
1
374
The embodiment of Google. gone like the darkness at dawn.
1
78
With what mechanism? Would this effectively not be the end of HuggingFace?
David Shapiro (L/0)
@DaveShapi
Jun 27
Do you think America will make open source AI illegal?
107
You can teach a biologist AI, but can you teach an AI scientist biology?
1
1
163
Load more