App 介绍

Uni-Fold 是深势科技于2021年推出的高精度蛋白质结构预测模型。在 Google DeepMind 推出的 AlphaFold [1] 基础上，深势科技团队成功突破技术壁垒，使用 PyTorch 复现了 AlphaFold2 及其完整训练过程，并从训练数据、模型结构、代码实现等多方面进行了优化与改进。经过不断迭代，目前 Uni-Fold 已达到预测精度超越 AlphaFold、推理速度显著提升、端到端训练速度提升超1倍的成就 [2]。目前的 Uni-Fold 支持蛋白质单体与复合物的高精度结构预测。同时，对于具有旋转对称性的蛋白质同源高聚体，Uni-Fold 提供预测方案 UF-Symmetry [3]，使结构预测复杂度与聚体数目无关，从数量级上提升了此类任务的预测速度。此外，Uni-Fold@Apps 采用MSA搜索方案 MMSeqs2，极大的提升了同源序列搜索的效率。对于一般的预测任务，Uni-Fold@Apps 在数分钟左右即可返回预测结果。

参考文献

[1] J Jumper et al. Highly Accurate Protein Structure Prediction with AlphaFold. Nature, 2021. doi: https://doi.org/10.1038/s41586-021-03819-2

[2] Z Li et al. Uni-Fold: An Open-Source Platform for Developing Protein Folding Models beyond AlphaFold. In bioRxiv, 2022. doi: https://doi.org/10.1101/2022.08.04.502811

[3] Z Li et al. Uni-Fold Symmetry: Harnessing Symmetry in Folding Large Protein Complexes. In bioRxiv, 2022. doi: https://doi.org/10.1101/2022.08.30.505833

引用格式

Uni-Fold:

@article {uni-fold,
	author = {Li, Ziyao and Liu, Xuyang and Chen, Weijie and Shen, Fan and Bi, Hangrui and Ke, Guolin and Zhang, Linfeng},
	title = {Uni-Fold: An Open-Source Platform for Developing Protein Folding Models beyond AlphaFold},
	year = {2022},
	doi = {10.1101/2022.08.04.502811},
	URL = {https://www.biorxiv.org/content/10.1101/2022.08.04.502811v3},
	eprint = {https://www.biorxiv.org/content/10.1101/2022.08.04.502811v3.full.pdf},
	journal = {bioRxiv}
}

UF-Symmetry:

@article {uf-symmetry,
	author = {Li, Ziyao and Yang, Shuwen and Liu, Xuyang and Chen, Weijie and Wen, Han and Shen, Fan and Ke, Guolin and Zhang, Linfeng},
	title = {Uni-Fold Symmetry: Harnessing Symmetry in Folding Large Protein Complexes},
	year = {2022},
	doi = {10.1101/2022.08.30.505833},
	URL = {https://www.biorxiv.org/content/early/2022/08/30/2022.08.30.505833},
	eprint = {https://www.biorxiv.org/content/early/2022/08/30/2022.08.30.505833.full.pdf},
	journal = {bioRxiv}
}