Today we release Rhaister, an elegant statistical model that predicts drug phenotypes in new contexts w/ accuracies comparable to experimental assays. And dropping Emerald Bay, a 2M cell dataset measuring long time-course phenotypes across 1000s of drug-cell line interactions.

I have two tickets for the Bob Dylan concert in Berkeley theater tonight. Can no longer go. Dm me if interested.

The government should not be regulating AI to this extent. Not like this. I’ve been against onerous regs when Anthropic and the safety community was pushing for it. And I’m against it now that they got what they asked for.

So let’s say I build an application that calls fable API in the backend. Can a foreign national use that application?

Our perturbation prediction manuscript is now on bioRxiv, for aficionados of DOIs: biorxiv.org/content/10.64898…

I want to see how controversial this is. The statement @elonmusk is the greatest entrepreneur since 1800 is

wow. what clarity "So, we may ask two things of any kind of work. Is its correctness private and expensive to establish, the kind of truth that exists only inside someone's data? And is it walled off, locked inside a system you can't get into? Set those against how saturated the task is, and you get a 2x2. Saturated work with public answers is commodity tokens, and open models own it. Frontier work with public answers, where coding benchmarks live, is where the labs win, because when the eval is free, owning it counts for nothing. The prize is the last corner, the untrainable one: frontier work whose correctness exists only in private."

A very thoughtful account of how we got here by @vallens who led the work on Rhaister.

Good to see folks going back to the basics. Been saying this for a while, that the whole AI virtual cell space has been putting parameters first & expt design second (or not at all) when the latter is the most important to match your questions of interest. 1/

Proud to see our ML team's work go public: Rhaister Emerald Bay deliver fast, interpretable drug-response prediction across diverse cancer contexts; matching far more expensive models, and openly available for everyone to build on. @tahoe_ai @shreshth_gandhi @vallens Read more Paper: tahoebio-assets.com/rhaister… Model and datasets: huggingface.co/collections/t…

We went back to basics to model cell states. Introducing our Rhaister model Emerald Bay dataset. See link in comments.

Congratulations @tahoe_ai team! A reminder that perturbation biology still rewards the basics: the right statistical framing and the right data. And thanks for Emerald Bay... already scrambling to add it to our current projects :D

amazing work! turns out predicting high-level cell features (used as benchmarks of virtual cell models) doesn't require complex models—a key insight into teaching machines biology

@tahoe_ai team doing what they do best! A key takeaway from VCC’25 for us was that in the current data regimes, elegant statistical models match if not surpass far more complex transformer-based architectures! “Going back to the basics” is such a great title for this… to say nothing of the 2M cell dataset! Kudos…

Huge improvement over SOTA virtual cell evals with a CPU based model!

At Tahoe, one of the reasons we are excited about perturbation prediction is to ultimately predict how patient samples will respond to any of the many drugs in our library