DreamFold: A World Model to efficiently generate protein folding pathways in the latent space
Alan Ianeselli, Jewon Im, Eddie Cavallin, Mark B. Gerstein · 2026
Read DreamFold: A World Model to efficiently generate protein folding pathways in the latent space by Alan Ianeselli, Jewon Im, Eddie Cavallin, Mark B. Gerstein (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13625212.txt.
Description
Alan Ianeselli and co-authors build a world model that imagines how proteins fold, in work from 2026. Still photos of protein shapes are plentiful, but movies of folding are rare. AI models need folding paths, not just end shapes, to learn dynamics.
Their system, called DreamFold, first compresses shapes and angle moves into short codes. A second network then predicts the next code after each move, forming a fast inner world. Inside that world, a learning agent practices steering folds toward a target shape. Decoded paths give full atomic movies for study. The approach borrows a winning idea from robotics to speed protein research.
Catalog record
| CREATOR | Alan Ianeselli, Jewon Im, Eddie Cavallin, Mark B. Gerstein |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |