用简单乘法替代三角注意力，大幅加速蛋白质结构预测。
Triangle Multiplication Is All You Need For Biomolecular Structure Representations
- 用三角乘法替代三角注意力，简化几何推理机制。
- 长序列推理速度提升4倍，训练成本降低34%。
- 适合大规模蛋白复合物与药物筛选场景。
AlphaFold 改变了蛋白质结构预测，但虚拟配体筛选、全蛋白组折叠和从头设计等新应用对大规模预测提出需求，导致运行时间和内存开销成为瓶颈。核心问题在于 AlphaFold3 类模型的 Pairformer 架构依赖计算昂贵的三角形原语——尤其是三角注意力——进行成对推理。我们提出 Pairmixer，一种简化替代方案，移除了三角注意力，同时保留了对结构预测至关重要的高阶几何推理能力。Pairmixer 显著提升计算效率，在折叠与对接基准上达到当前最优水平，长序列推理速度最高提升4倍，训练成本降低34%。其高效性缓解了下游应用的计算负担，如大蛋白复合物建模、高通量配体与结合剂筛选及基于幻觉的设计。在 BoltzDesign 中，Pairmixer 实现超过2倍更快采样，并支持比 Pairformer 内存限制长30%的序列。代码已开源：https://github.com/genesistherapeutics/pairmixer。
原文摘要 · Abstract (English)
AlphaFold has transformed protein structure prediction, but emerging applications such as virtual ligand screening, proteome-wide folding, and de novo binder design demand predictions at a massive scale, where runtime and memory costs become prohibitive. A major bottleneck lies in the Pairformer backbone of AlphaFold3-style models, which relies on computationally expensive triangular primitives-especially triangle attention-for pairwise reasoning. We introduce Pairmixer, a streamlined alternative that eliminates triangle attention while preserving higher-order geometric reasoning capabilities that are critical for structure prediction. Pairmixer substantially improves computational efficiency, matching state-of-the-art structure predictors across folding and docking benchmarks, delivering up to 4x faster inference on long sequences while reducing training cost by 34%. Its efficiency alleviates the computational burden of downstream applications such as modeling large protein complexes, high-throughput ligand and binder screening, and hallucination-based design. Within BoltzDesign, for example, Pairmixer delivers over 2x faster sampling and scales to sequences ~30% longer than the memory limits of Pairformer. Code is available at https://github.com/genesistherapeutics/pairmixer.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。