A working fluorescent protein generated by a language model
ESM3 designed esmGFP, a bright green fluorescent protein sharing only 58% of its sequence with the closest known natural fluorescent protein, a distance the authors estimate at over 500 million years of natural evolution.
- Model
- ESM3 (98B)
- Field
- Biology
- Date
- 2025-01-16
- Human collaborators
- Alexander Rives, Tom Sercu
Sources
Original work
Announcement
What was found
ESM3 is a masked language model over protein sequence, structure and function jointly. Prompted with the residues that form the GFP chromophore and little else, and run through iterative generation, it produced candidates that were synthesised and assayed in the lab. esmGFP fluoresces at brightness comparable to natural GFPs while differing at 96 of 229 positions from its nearest natural relative. Published in Science.
Novelty check
Engineered GFP variants are routine but sit a handful of mutations from a natural template, and directed evolution moves in small steps. A functional fluorescent protein this far from any known sequence had not been reported, and the paper includes the expression, spectra and brightness measurements rather than a prediction alone.
Caveats and known objections
The 500-million-year figure is an inference from a sequence-divergence-to-time calibration, not a measurement, and has been criticised as a marketing frame. esmGFP was one hit from a generated and screened set; the paper reports the funnel. The design goal and the chromophore constraint were supplied by humans. No independent re-synthesis of esmGFP by an unaffiliated lab was on record when this entry was written.
Nobody outside the lab has checked this yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Peer reviewed and AI-led.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded peer reviewed for verification and ai-led for autonomy. What these mean.
Cite this entry
whataifound.org. (2025). A working fluorescent protein generated by a language model. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2025-01-16-esmgfpBibTeX
@misc{whataifound-evolutionaryscale-2025-esmgfp,
title = {A working fluorescent protein generated by a language model},
author = {{whataifound.org}},
year = {2025},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by EvolutionaryScale. Verification: Peer reviewed. Autonomy: AI-led.},
url = {https://whataifound.org/finding/2025-01-16-esmgfp}
}