| 19 | 0 | 95 |
| 下载次数 | 被引频次 | 阅读次数 |
在新药研发过程中,由于缺乏历史相互作用数据,如何准确识别冷启动药物的潜在风险成为关键挑战之一。针对现有预测模型在冷启动场景下泛化能力不足、分子结构细粒度信息建模不充分以及多模态信息融合能力有限等问题,本文提出一种多尺度键感知与多模态融合的冷启动药物相互作用(drug-drug interaction, DDI)风险识别模型。该模型构建键感知分子图学习模块,在多尺度上刻画化学键与子结构间复杂关系,运用冷启动泛化模块增强具有功能语义的新药数据,融合分子结构信息、多源属性信息及功能语义特征,实现多模态联合建模。实验结果表明,在冷启动场景下,本文模型在准确率(accuracy, ACC)、受试者工作特征曲线下面积(area under the receiver operating characteristic curve, AUC)及f1得分等指标总体优于已有预测模型,消融实验、参数敏感性分析和案例研究也表明本文模型具有更强的泛化能力、稳定性和有效性。本研究工作成果为新药早期安全性评估及用药风险识别提供一种计算工具。
Abstract：In the context of new drug development, the lack of historical interaction data makes the accurate identification of potential risks for cold-start drugs one of the significant challenges. To address the limitations of existing methods regarding insufficient generalization capability in cold-start scenarios, inadequate modeling of fine-grained molecular structural information, and limited multi-modal fusion capacity, this paper proposes a novel approach called multi-scale bond-aware and multi-modal fusion for risk identification of drug-drug interaction in cold start scenario. The proposed method constructs a bond-aware molecular graph learning module to characterize the complex relationships between chemical bonds and substructures at multiple scales, employs a cold start generalization module to augment new drugs data with functional semantics, and integrates molecular structural information, multi-source attributes and functional semantic features to achieve multi-modal joint modeling. Experimental results demonstrate that in cold-start scenario, the proposed method generally outperforms existing baseline methods across multiple evaluation metrics including accuracy( ACC), area under the receiver operating characteristic curve( AUC), and f1-score, and the ablation experiment, parameter sensitivity, and case study also demonstrate that the proposed model exhibits stronger generalization ability, stability, and effectiveness. This study will provide a computational tool for the early-stage safety assessment of new drugs and potential medication risk identification.
杜丽娜,2004.格列吡嗪[J].临床药物治疗杂志,2(4):40-46.[Du L N,2004.Glipizide[J].Clinical Medication Journal,2(4):40-46.]
顾大伟,李璐,2023.住院肿瘤患者潜在药物相互作用的发生率和严重程度及危险因素分析[J].现代肿瘤医学,31(17):3270-3275.[Gu D W,Li L,2023.Analysis of the incidence,severity and risk factors of potential drug interactions in hospitalized cancer patients[J].Journal of Modern Oncology,31(17):3270-3275.]
郭文博,龙伟,蒋林华,等,2024.基于图神经网络的药物相互作用预测方法研究综述[J].现代计算机,30(20):36-40.[Guo W B,Long W,Jiang L H,et al.,2024.A review of prediction methods of drug interaction based on graph neural network[J].Modern Computer,30(20):36-40.]
胡鹏,阚红星,王永康,等,2023.基于异构图注意力网络的药物相互作用预测[J].中国医疗设备,38(5):110-114.[Hu P,Kan H X,Wang Y K,et al.,2023.Drug interaction prediction based on heterogeneous graph attention network[J].China Medical Devices,38(5):110-114.]
康雁,徐玉龙,寇勇奇,等,2022.基于Transformer和LSTM的药物相互作用预测[J].计算机科学,49(S1):17-21.[Kang Y,Xu Y L,Kou Y Q,et al.,2022.Drug-drug interaction prediction based on transformer and LSTM[J].Computer Science,49(S1):17-21.]
刘文潇,2023.基于深度学习的药物相互作用预测算法研究[D].上海:东华大学.[Liu W X,2023.Research on deep learning-based algorithms for drug-drug interaction prediction,[D].Shanghai:Donghua University.]
孙搏,付淑军,陈桂良,等,2021.药物相互作用研究在新药研发和审评决策中的应用[J].中国临床药理学与治疗学,26(10):1095-1102.[Sun B,Fu S J,Chen G L,et al.,2021.Applications of drug interaction study in new drug development and regulatory decision-making[J].Chinese Journal of Clinical Pharmacology and Therapeutics,26(10):1095-1102.]
夏文凯,孙成磊,刘新灵,等,2017.α-糖苷酶抑制剂米格列醇的研究进展[J].药品评价,14(13):9-12.[Xia W K,Sun C L,Liu X L,et al.,2017.Progress ofα-glycosidase inhibitor miglitol[J].Drug Evaluation,14(13):9-12.]
杨霞,韩春燕,琚生根,2022.基于实体信息和图神经网络的药物相互作用关系抽取[J].四川大学学报(自然科学版),59(2):42-50.[Yang X,Han C Y,Ju S G,2022.Drug-Drug relationship extraction based on entity information and graph neural networks[J].Journal of Sichuan University(Natural Science Edition),59(2):42-50.]
张立强,秦玉芳,陈明,2023.SMILESynergy:基于Transformer预训练模型的抗癌药物协同作用预测[J].生物医学工程学杂志,40(3):544-551.[Zhang L Q,Qin Y F,Chen M,2023.SMILESynergy:anticancer drug synergy prediction based on Transformer pre-trained model[J].Journal of Biomedical Engineering,40(3):544-551.]
Abbas K,Hao C,Yong X,et al.,2025.Graph neural networkbased drug-drug interaction prediction[J].Sci.Rep.,15(1):30340.
Chen Q Y,Li X,Geng K N,et al.,2023.Context-aware safe medication recommendations with molecular graph and DDIgraph embedding[J].Proc.AAAI Conf.Artif.Intell.,37(6):7053-7060.
Deng Y F,Xu X R,Qiu Y,et al.,2020.A multimodal deep learning framework for predicting drug-drug interaction events[J].Bioinform.(Oxf.Engl.),36(15):4316-4322.
Feng Y H,Zhang S W,Shi J Y,2020.DPDDI:a deep predictor for drug-drug interactions[J].BMC Bioinform.,21(1):419.
Feng Y H,Zhang S W,Zhang Q Q,et al.,2022.deep MDDI:a deep graph convolutional network framework for multi-label prediction of drug-drug interactions[J].Anal.Biochem.,646:114631.
Fitzmaurice M G,Wong A,Akerberg H,et al.,2019.Evaluation of potential drug-drug interactions in adults in the intensive care unit:a systematic review and meta-analysis[J].Drug Saf.,42(9):1035-1044.
Goulooze S C,Cohen A F,Rissmann R,2015.Mirabegron[J].Br.J.Clin.Pharmacol.,80(4):762-764.
Lee G,Park C,Ahn J,2019.Novel deep learning model for more accurate prediction of drug-drug interaction effects[J].BMC Bioinform.,20(1):415.
Lin S G,Chen W Z,Chen G W,et al.,2022a.MDDI-SCL:predicting multi-type drug-drug interactions via supervised contrastive learning[J].J.Cheminf.,14(1):81.
Lin S G,Wang Y J,Zhang L F,et al.,2022b.MDF-SA-DDI:predicting drug-drug interaction events based on multi-source drug fusion,multi-source feature fusion and transformer selfattention mechanism[J].Brief.Bioinform.,23 (1):bbab421.
Lin S X,Cui G D,Tang C,et al.,2025.ATLAS-DMPNN:an attention-guided topological framework for enhanced ADMETproperty prediction//2025 International Conference on New Trends in Computational Intelligence (NTCI).IEEE:142-147.
Liu Z,Wang X N,Yu H,et al.,2022.Predict multi-type drugdrug interactions in cold start scenario[J].BMC Bioinform.,23(1):75.
Lyu T F,Gao J L,Tian L,et al.,2021.MDNN:a multimodal deep neural network for predicting drug-drug interaction events//Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence.International Joint Conferences on Artificial Intelligence Organization:3536-3542.
Niu D J,Zhang L W,Zhang B Y,et al.,2024a.DAS-DDI:a dual-view framework with drug association and drug structure for drug-drug interaction prediction[J].J.Biomed.Inform.,156:104672.
Niu D J,Xu L,Pan S R,et al.,2024b.SRR-DDI:a drug-drug interaction prediction model with substructure refined representation learning based on self-attention mechanism[J].Knowl.Based Syst.,285:111337.
Nusair M B,Al-Azzam S I,Arabyat R M,et al.,2020.The prevalence and severity of potential drug-drug interactions among adult polypharmacy patients at outpatient clinics in Jordan[J].Saudi Pharm.J.,28(2):155-160.
Nyamabo A K,Yu H,Shi J Y,2021.SSI-DDI:substructuresubstructure interactions for drug-drug interaction prediction[J].Brief.Bioinform.,22(6):bbab133.
Nyamabo A K,Yu H,Liu Z,et al.,2022.Drug-drug interaction prediction with learnable size-adaptive molecular substructures[J].Brief.Bioinform.,23(1):bbab441.
Pang Y,Chen Y H,Lin M J,et al.,2024.MMSyn:a new multimodal deep learning framework for enhanced prediction of synergistic drug combinations[J].J.Chem.Inf.Model.,64(9):3689-3705.
Peng L H,Liu X,Chen M,et al.,2024.MGNDTI:a drugtarget interaction prediction framework based on multimodal representation learning and the gating mechanism[J].J.Chem.Inf.Model.,64(16):6684-6698.
Pharmacia&Upjohn Company,2012.GlysetⓇ(miglitol) tablets for oral use[EB/OL].U.S.Food and Drug Administration,(2012-09-17)[2026-05-28].https://www.accessdata.fda.gov/drugsatfda_docs/label/2012/020682s010lbl.pdf.
Ryu J Y,Kim H U,Lee S Y,2018.Deep learning improves prediction of drug-drug and drug-food interactions[J].Proc.Natl.Acad.Sci.USA,115(18):E4304-E4311.
Tang Z C,Chen G X,Yang H L,et al.,2024.DSIL-DDI:a domain-invariant substructure interaction learning for generalizable drug-drug interaction prediction[J].IEEE Trans.Neural Netw.Learn.Syst.,35(8):10552-10560.
Wang H C,Lian D F,Zhang Y,et al.,2020.Go GNN:graph of graphs neural network for predicting structured entity interactions[PP/OL].ar Xiv (2020-05-12)[2026-05-29].https://doi.org/10.48550/ar Xiv.2005.05537.
Wang Y F,Zhou L,Dutreix C,et al.,2008.Effects of imatinib(Glivec) on the pharmacokinetics of metoprolol,a CYP2D6substrate,in Chinese patients with chronic myelogenous leukaemia[J].Br.J.Clin.Pharmacol.,65(6):885-892.
Weininger D,1988.SMILES,a chemical language and information system.1.Introduction to methodology and encoding rules[J].J.Chem.Inf.Comput.Sci.,28(1):31-36.
Wu D,Sun W,He Y,et al.,2024.MKG-FENN:a multimodal knowledge graph fused end-to-end neural network for accurate drug-drug interaction prediction[J].Proc.AAAI Conf.Artif.Intell.,38(9):10216-10224.
Xiong Z K,Liu S C,Huang F,et al.,2023.Multi-relational contrastive learning graph neural network for drug-drug interaction event prediction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.Washington:AAAI Press,37(4):5339-5347.
Xu N,Wang P H,Chen L,et al.,2019.MR-GNN:multiresolution and dual graph neural network for predicting structured entity interactions[C]//Proceedings of the TwentyEighth International Joint Conference on Artificial Intelligence.Macao:IJCAI,2019:3986-3992.
Yang Z D,Zhong W H,Lv Q J,et al.,2022.Learning sizeadaptive molecular substructures for explainable drug-drug interaction prediction by substructure-aware graph neural network[J].Chem.Sci.,13(29):8693-8703.
Yu H,Wang J,Zhao S Y,et al.,2024.GGI-DDI:identification for key molecular substructures by granule learning to interpret predicted drug-drug interactions[J].Expert Syst.Appl.,240:122500.
Yu Y,Huang K X,Zhang C,et al.,2021.Sum GNN:multityped drug interaction prediction via efficient knowledge graph summarization[J].Bioinformatics,37 (18):2988-2995.
Yuan Y N,Yue J Q,Zhang R S,et al.,2025.PHGL-DDI:a pre-training based hierarchical graph learning framework for drug-drug interaction prediction[J].Expert Syst.Appl.,270:126408.
Yun S,Jeong M,Kim R,et al.,2020.Graph transformer networks[PP/OL].V2.ar Xiv (2020-02-05)[2026-05-29].https://doi.org/10.48550/ar Xiv.1911.06455.
Zhang J X,Zeng Y,Tang C Y,et al.,2025.Contrastive learning-based multi-mechanism disentangled assessment for drugdrug interaction[J].BMC Bioinform.,26(1):286.
Zhong Y,Zheng H B,Chen X M,et al.,2023.DDI-GCN:Drug-drug interaction prediction via explainable graph convolutional networks[J].Artif.Intell.Med.,144:102640.
Zitnik M,Agrawal M,Leskovec J,2018.Modeling polypharmacy side effects with graph convolutional networks[J].Bioinform.(Oxf.Engl.),34(13):i457-i466.
基本信息:
DOI：10.13417/j.gab.045.000810
中图分类号:TP18;R9
引用信息:
[1]曾雲鞠,张锦雄,王巫亮,等.基于多尺度键感知与多模态融合的冷启动药物潜在风险识别[J].基因组学与应用生物学,2026,45(04):810-827.DOI:10.13417/j.gab.045.000810.
基金信息:
国家自然科学基金项目(62362004)资助
2026-08-25
2026-08-25