
info@juzhikan.asia
重庆师范大学,重庆市,401331;
摘要:药物-靶点相互作用(DTI)预测是计算机辅助药物发现的核心环节,但现有方法常受限于多模态特征融合不充分及跨域冷启动泛化能力弱。为此,本文提出一种基于分层混合专家与多模态融合深度学习框架(HMDF-DTI)。该模型构建了双通道编码机制,自适应提取分子的空间拓扑与序列语义,同时依托双路由分层混合专家(MoE)架构实现递进式跨模态交互,并引入对抗正则化约束提取领域无关特征。实验表明,该框架在三大基准数据集上的性能均显著优于对比模型,特别是在严苛的冷启动场景下展现出卓越的跨域泛化优势。
关键词:药物-靶点相互作用;多模态特征融合;冷启动;混合专家模型;Transformer网络
参考文献
[1]Xu K, Ding Y, Hou S, et al. Domain adaptive and fine-grained anomaly detection for single-cell sequencing data and beyond [C]. Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-24), 2024: 6125-6133.
[2]Jumper J, Evans R, Pritzel A, et al. Highly accurate protein structure prediction with alphafold [J]. Nature, 2021, 596(7873): 583-589.
[3]Jumper J, Evans R, Pritzel K, et al. Highly accurate protein structure prediction with AlphaFold [J]. Nature, 2021, 596(7873): 583-589.
[4]Shazeer N, Mirhoseini A, Maziarz K, et al. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer [C]. International Conference on Learning Representations (ICLR), 2017.