A good prediction is useful. But it rarely tells you what to do next.
You still want to know whether another model sees the same fold. Whether a high-ranking construct is likely to express, or actually behave downstream. Whether the protocol can be run in the lab you have, not the ideal lab the method assumes.
That is what we focused on this month. Three updates are now live in Orbion:
- Boltz-2 and ESMFold — on-demand structure prediction as a second opinion
- Split Design scores — Yield and Developability, separated
- Lab-aware Bench — numbered protocols that respect your equipment
Boltz-2 and ESMFold
While AlphaFold2 and AlphaFold2-Multimer are the default models for structure prediction on the platform, you can now run Boltz-2 or ESMFold on any target, on demand, inside Characterize, Stabilize, and Design.
- ESMFold — single-sequence folding, no MSA, in seconds
- Boltz-2 — structure prediction with affinity estimates where applicable
- See an ESMFold prediction on a construct you've designed
Use them as a second opinion before committing to a design path.
Yield and Developability on Design
Design used to combine expression and developability into one score. But they answer different questions.
- Yield — is this construct likely to express well?
- Developability — will it stay soluble, stable, and well-behaved downstream?
- A construct can win on one and lose on the other — now you see which
Rank by the property the campaign actually turns on.
Bench is now aware of your lab
Bench protocols now read as numbered, step-by-step procedures — and respect what your lab can actually do.
- Step-by-step view: every protocol as a numbered procedure, not a wall of text
- Add your lab's limitations — no insect cell line, no AKTA, no −80 °C — and protocols adapt around them
- Get a procedure the bench can run on day one, not after a rewrite
Tell Bench what you've got and it plans for that lab. You can set your lab equipment under Organization Settings.
What's the one thing we should ship next? Open your dashboard — and tell us. We read every reply.
