Illuminating protein space with a programmable generative model
Problem
The space of possible protein molecules is vastly larger than the space of functional ones, making it hard for computation and experiments to access proteins with desired properties and functions.
Why it matters
Designing novel proteins could unlock new medicines, materials and functions beyond those found in nature.
What the AI did
Chroma, a generative model using a polymer-aware diffusion process and sub-quadratic neural architecture, directly samples novel protein structures and sequences, steerable towards desired properties via conditioning.
Results
not stated in source
Impact
Peer-reviewed Nature paper presenting a programmable generative model for protein design; abstract shows capabilities but no measured outcomes.
Record
| Venue | Nature |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | John Ingraham, Max Baranov, Zak Costello, Karl W. Barber, Wujie Wang, Ahmed Ismail, Vincent Frappier, Dana M. Lord et al. |
| Citations | 432 (OpenAlex, at import time) |
| References | 54 |
| Model or method | Chroma (diffusion-based generative model) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41586-023-06728-8 |
| Open access | hybrid |
| Keywords | protein design, generative model, diffusion, protein structures, Bayesian inference |
Importance rating
major — millions of lives, an industry or a field
Protein design underpins potential new medicines, materials and functions beyond nature; abstract frames it as accessing a vastly larger functional protein space, a foundational capability.