Flows, straight but not so fast: Exploring the design space of Rectified Flows in Protein Design
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909874988777472 |
|---|---|
| author | Chen, Junhua Mathis, Simon Harris, Charles Didi, Kieran Lio, Pietro |
| author_facet | Chen, Junhua Mathis, Simon Harris, Charles Didi, Kieran Lio, Pietro |
| contents | Generative modeling techniques such as Diffusion and Flow Matching have achieved significant successes in generating designable and diverse protein backbones. However, many current models are computationally expensive, requiring hundreds or even thousands of function evaluations (NFEs) to yield samples of acceptable quality, which can become a bottleneck in practical design campaigns that often generate $10^4\ -\ 10^6$ designs per target. In image generation, Rectified Flows (ReFlow) can significantly reduce the required NFEs for a given target quality, but their application in protein backbone generation has been less studied. We apply ReFlow to improve the low NFE performance of pretrained SE(3) flow matching models for protein backbone generation and systematically study ReFlow design choices in the context of protein generation in data curation, training and inference time settings. In particular, we (1) show that ReFlow in the protein domain is particularly sensitive to the choice of coupling generation and annealing, (2) demonstrate how useful design choices for ReFlow in the image domain do not directly translate to better performance on proteins, and (3) make improvements to ReFlow methodology for proteins. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_24732 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Flows, straight but not so fast: Exploring the design space of Rectified Flows in Protein Design Chen, Junhua Mathis, Simon Harris, Charles Didi, Kieran Lio, Pietro Biomolecules Artificial Intelligence Generative modeling techniques such as Diffusion and Flow Matching have achieved significant successes in generating designable and diverse protein backbones. However, many current models are computationally expensive, requiring hundreds or even thousands of function evaluations (NFEs) to yield samples of acceptable quality, which can become a bottleneck in practical design campaigns that often generate $10^4\ -\ 10^6$ designs per target. In image generation, Rectified Flows (ReFlow) can significantly reduce the required NFEs for a given target quality, but their application in protein backbone generation has been less studied. We apply ReFlow to improve the low NFE performance of pretrained SE(3) flow matching models for protein backbone generation and systematically study ReFlow design choices in the context of protein generation in data curation, training and inference time settings. In particular, we (1) show that ReFlow in the protein domain is particularly sensitive to the choice of coupling generation and annealing, (2) demonstrate how useful design choices for ReFlow in the image domain do not directly translate to better performance on proteins, and (3) make improvements to ReFlow methodology for proteins. |
| title | Flows, straight but not so fast: Exploring the design space of Rectified Flows in Protein Design |
| topic | Biomolecules Artificial Intelligence |
| url | https://arxiv.org/abs/2510.24732 |