ML research
OpenDDE releases an open-source co-folding drug-discovery model
- Category: Paper
- Status: developing
- Sources: arXiv 2607.03787
- Summary: A preprint (v1) introduces the Open Drug Discovery Engine (OpenDDE), described as an open-source, all-atom biomolecular foundation model that uses co-folding as the entry point to structure prediction across biomolecular complexes and as a shared layer for de novo design, affinity estimation, and structure-conditioned optimization. The authors report reaching IsoDDE-level co-folding accuracy within a reproducible, openly accessible pipeline that combines all-atom architecture, atomic latent reasoning, inference optimization, and large-scale data processing. The accuracy claims are the project's own and are not independently reproduced.
- Why it matters: Open all-atom co-folding tooling extends the AlphaFold, Boltz, and Chai line of open structure-prediction models that practitioners can run and adapt without proprietary access.