EC-Conf: An Ultra-fast Diffusion Model for Molecular Conformation Generation with Equivariant Consistency

Zhiguang Fan,Yuedong Yang,Mingyuan Xu,Hongming Chen

2023-11-23

Abstract:Despite recent advancement in 3D molecule conformation generation driven by diffusion models, its high computational cost in iterative diffusion/denoising process limits its application. In this paper, an equivariant consistency model (EC-Conf) was proposed as a fast diffusion method for low-energy conformation generation. In EC-Conf, a modified SE (3)-equivariant transformer model was directly used to encode the Cartesian molecular conformations and a highly efficient consistency diffusion process was carried out to generate molecular conformations. It was demonstrated that, with only one sampling step, it can already achieve comparable quality to other diffusion-based models running with thousands denoising steps. Its performance can be further improved with a few more sampling iterations. The performance of EC-Conf is evaluated on both GEOM-QM9 and GEOM-Drugs sets. Our results demonstrate that the efficiency of EC-Conf for learning the distribution of low energy molecular conformation is at least two magnitudes higher than current SOTA diffusion models and could potentially become a useful tool for conformation generation and sampling. We release our code at <a class="link-external link-https" href="https://github.com/zhi520/EcConf" rel="external noopener nofollow">this https URL</a>.

Biomolecules,Chemical Physics

What problem does this paper attempt to address?

The paper aims to address the high computational cost associated with molecular conformation generation, particularly during the iterative diffusion/denoising process. Specifically, existing diffusion model-based methods require numerous iterative steps to generate low-energy molecular conformations, which limits their applicability. To this end, the authors propose a new Equivariant Consistency Model (EC-Conf), an ultra-fast diffusion method for generating low-energy molecular conformations.

### Problems the paper attempts to solve:
1. **Improve computational efficiency**: Existing diffusion model-based methods require thousands of iterations to generate high-quality molecular conformations, whereas EC-Conf can achieve similar results with just one sampling or a few iterations.
2. **Maintain generation quality**: Despite the significant improvement in computational efficiency, EC-Conf is still able to generate high-quality low-energy molecular conformations, with performance comparable to the current state-of-the-art diffusion models.
3. **Equivariance**: Ensure that the generated molecular conformations possess SE(3) equivariance, meaning that translation and rotation operations in three-dimensional space do not affect their scalar properties (such as energy or ADMET properties).

With these improvements, EC-Conf significantly enhances computational efficiency while maintaining generation quality, making it a promising tool for molecular conformation generation and sampling.