发表机构
I2A institute, TÉLUQ University; LIO, CRCHUM; LCSI, ESI; CoMeT UR 24344, Universite de Poitiers; University of Ottawa; Lebanese American University（I2A研究所，TÉLUQ大学; LIO，蒙特利尔大学医院研究中心; LCSI，阿尔及尔国立高等信息学院; CoMeT UR 24344，普瓦捷大学; 渥太华大学; 黎巴嫩美国大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出一种仅基于序列的DTI预测架构，结合ChemBERTa和ESM-2预训练模型与层级交叉注意力，在多个基准上取得优异性能，仅用2520万参数。
AI 中文摘要
预测药物-靶点相互作用（DTIs）是计算药物发现中的核心任务，在虚拟筛选、药物重定位和治疗候选物优先级排序中具有直接应用。尽管最近的深度学习方法改进了DTI预测，但许多基于序列的模型仍独立处理药物和蛋白质，仅在预测后期组合其表示。这限制了它们显式建模化学子结构与蛋白质序列区域之间跨分子依赖性的能力。在本文中，我们提出了一种仅基于序列的DTI预测架构，该架构结合了两个预训练语言模型——用于药物SMILES字符串的ChemBERTa和用于蛋白质氨基酸序列的ESM-2，以及一个层级交互模块。所提出的模型首先使用预训练编码器提取上下文表示，然后应用一维卷积层压缩局部序列模式，接着采用受分子识别诱导契合观点启发的序列双向交叉注意力机制。最后，基于注意力的池化构建固定大小的交互感知向量以进行二元预测。在BIOSNAP、Davis和BindingDB上的实验表明，所提出的模型在BIOSNAP上取得了最佳性能，在Davis上匹配了最佳AUROC，并在BindingDB上保持竞争力，同时仅使用2520万个可训练参数。消融结果证实了CNN和交叉注意力模块的贡献，冷启动实验表明对未见蛋白质和药物具有良好的泛化能力。
英文摘要
Predicting Drug-Target Interactions~(DTIs) is a central task in computational drug discovery, with direct applications in virtual screening, drug repurposing, and therapeutic candidate prioritization. Although recent deep learning methods have improved DTI prediction, many sequence-based models still process drugs and proteins independently and only combine their representations at a late prediction stage. This limits their ability to explicitly model cross-molecular dependencies between chemical substructures and protein sequence regions. In this paper, we propose a sequence-only DTI prediction architecture that combines two pre-trained language models, ChemBERTa for drug SMILES strings and ESM-2 for protein amino acid sequences, with a hierarchical interaction module. The proposed model first extracts contextual representations using pre-trained encoders, then applies 1D convolutional layers to condense local sequence patterns, followed by a sequential bidirectional cross-attention mechanism inspired by the induced-fit view of molecular recognition. Finally, attention-based pooling constructs fixed-size interaction-aware vectors for binary prediction. Experiments on BIOSNAP, Davis, and BindingDB show that the proposed model achieves the best performance on BIOSNAP, matches the best AUROC on Davis, and remains competitive on BindingDB while using only 25.2 million trainable parameters. Ablation results confirm the contribution of both the CNN and cross-attention modules, and cold-start experiments indicate promising generalization to unseen proteins and drugs.
