DCFold: Efficient Protein Structure Generation with Single Forward Pass
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909053921263616 |
|---|---|
| author | Zhang, Zhe Feng, Yuanning Song, Yuxuan Qiu, Keyue Zhou, Hao Ma, Wei-Ying |
| author_facet | Zhang, Zhe Feng, Yuanning Song, Yuxuan Qiu, Keyue Zhou, Hao Ma, Wei-Ying |
| contents | AlphaFold3 introduces a diffusion-based architecture that elevates protein structure prediction to all-atom resolution with improved accuracy. This state-of-the-art performance has established AlphaFold3 as a foundation model for diverse generation and design tasks. However, its iterative design substantially increases inference time, limiting practical deployment in downstream settings such as virtual screening and protein design. We propose DCFold, a single-step generative model that attains AlphaFold3-level accuracy. Our Dual Consistency training framework, which incorporates a novel Temporal Geodesic Matching (TGM) scheduler, enables DCFold to achieve a 15x acceleration in inference while maintaining predictive fidelity. We validate its effectiveness across both structure prediction and binder design benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_17899 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DCFold: Efficient Protein Structure Generation with Single Forward Pass Zhang, Zhe Feng, Yuanning Song, Yuxuan Qiu, Keyue Zhou, Hao Ma, Wei-Ying Machine Learning Artificial Intelligence Quantitative Methods AlphaFold3 introduces a diffusion-based architecture that elevates protein structure prediction to all-atom resolution with improved accuracy. This state-of-the-art performance has established AlphaFold3 as a foundation model for diverse generation and design tasks. However, its iterative design substantially increases inference time, limiting practical deployment in downstream settings such as virtual screening and protein design. We propose DCFold, a single-step generative model that attains AlphaFold3-level accuracy. Our Dual Consistency training framework, which incorporates a novel Temporal Geodesic Matching (TGM) scheduler, enables DCFold to achieve a 15x acceleration in inference while maintaining predictive fidelity. We validate its effectiveness across both structure prediction and binder design benchmarks. |
| title | DCFold: Efficient Protein Structure Generation with Single Forward Pass |
| topic | Machine Learning Artificial Intelligence Quantitative Methods |
| url | https://arxiv.org/abs/2605.17899 |