Independent validation · Nature Chemical Biology · 2026
The field just validated our approach — the hard way.
A peer-reviewed paper from a leading lab reached the same scientific conclusion behind our KEAP1 program. They got there by physically screening more than fifteen thousand peptides. We got there computationally — in two days.
Published · EPFL · Nat. Chem. Biol.
15,360
fully random cyclic peptides physically synthesized and screened to find an inhibitor of the Keap1–Nrf2 interaction — then refined through repeated design–build–test cycles, with X-ray crystal structures, into one membrane-permeable hit (peptide 30, 890 Da) active in living cells.
Peer-reviewed · wet-lab provenBiologic AI · KEAP1 program
2 days
to reach the same target and the same modality — novel, membrane-permeable cyclic peptides against Keap1–Nrf2 — by de novo computational design. 22 sequences, patent filed. A simpler construct we believe is more likely to work.
Designed in silicoJi et al., Generation of membrane-permeable cyclic peptides inhibiting protein–protein interaction, Nature Chemical Biology (2026). Their compound is wet-lab validated and peer-reviewed; our candidates are computationally designed and not yet wet-lab tested. The significance is the speed and capital efficiency of reaching the same scientific frontier.
What it means
Third-party proof of the science — and of our speed.
01 · The target is real
An independent group, publishing in a top journal, confirms Keap1–Nrf2 is a tractable, high-value target for cyclic peptides — the foundation of our lead program.
02 · The modality works
Their result shows membrane-permeable cyclic peptides can be built against this interaction from scratch — no known ligand or natural binding motif required.
03 · Our engine is fast
Where the field invested a major synthesis-and-screening campaign, our platform reached the same frontier computationally in two days — the capital-efficiency thesis, demonstrated.