ReQFlow: Rectified Quaternion Flow for Efficient and High-Quality Protein Backbone Generation
Angxiao Yue, Zichong Wang, Hongteng Xu
2025-02-24
Summary
This paper talks about ReQFlow, a new AI method for creating protein structures faster and more accurately than previous approaches. It uses a special way of representing protein shapes using mathematics to generate high-quality protein backbones efficiently.
What's the problem?
Creating new protein structures is important for biology and medicine, but current AI methods are often slow and don't always produce proteins that could exist in real life. This makes it hard to design new proteins for things like drug development or understanding diseases.
What's the solution?
The researchers developed ReQFlow, which uses a mathematical technique called quaternion flow to represent protein shapes. This method generates each part of the protein structure step by step, using both position and rotation information. They also made the process faster and more accurate by 'rectifying' or adjusting the model during use.
Why it matters?
This matters because ReQFlow can create protein structures much faster than other methods (37 times faster than one popular method and 62 times faster than another for a typical protein). It also produces more realistic protein structures. This could speed up drug discovery, help scientists understand diseases better, and lead to new biotechnology applications by making it easier to design custom proteins.
Abstract
Protein backbone generation plays a central role in de novo protein design and is significant for many biological and medical applications. Although diffusion and flow-based generative models provide potential solutions to this challenging task, they often generate proteins with undesired designability and suffer computational inefficiency. In this study, we propose a novel rectified quaternion flow (ReQFlow) matching method for fast and high-quality protein backbone generation. In particular, our method generates a local translation and a 3D rotation from random noise for each residue in a protein chain, which represents each 3D rotation as a unit quaternion and constructs its flow by spherical linear interpolation (SLERP) in an exponential format. We train the model by quaternion flow (QFlow) matching with guaranteed numerical stability and rectify the QFlow model to accelerate its inference and improve the designability of generated protein backbones, leading to the proposed ReQFlow model. Experiments show that ReQFlow achieves state-of-the-art performance in protein backbone generation while requiring much fewer sampling steps and significantly less inference time (e.g., being 37x faster than RFDiffusion and 62x faster than Genie2 when generating a backbone of length 300), demonstrating its effectiveness and efficiency. The code is available at https://github.com/AngxiaoYue/ReQFlow.