| 59 | 0 | 119 |
| 下载次数 | 被引频次 | 阅读次数 |
高效预测抗癌药物组合的协同性是制定个性化联合用药方案的关键。抗癌药物的生物活性高度依赖于其分子结构中的药效团,然而现有计算模型在深度挖掘与利用分子结构中的药效团特征方面仍存在局限。针对该问题,本研究提出一种基于多尺度结构相似性的抗癌药物组合协同性预测模型(multi-scale molecular structural similarity-based synergy prediction model for anticancer drug combinations, MSSynergy)。该模型利用三层图卷积神经网络对药物分子的多尺度结构特征进行分层表征,引入注意力机制量化药物在不同尺度下的结构相似度得分,从结构相似性的角度捕捉关键的抗肿瘤药效特征,并利用多层卷积神经网络提取癌细胞系的转录组基因表达特征,通过全连接层实现协同得分或分类的端到端预测。实验结果表明,MSSynergy模型的均方误差(mean squared error, MSE)、斯皮尔曼相关系数(Spearman correlation coefficient, SCC)与决定系数(coefficient of determination, R2)、受试者工作特征曲线下面积(area under the receiver operating characteristic curve, AUC)、精确率-召回率曲线下面积(area under the precision-recall curve, AUPR)及准确率(accuracy, ACC)分别达到0.010、 0.836、0.994、 0.940、 0.939和0.870,均高于已有方法,充分验证了MSSynergy模型能够为癌症精准治疗及临床药物筛选提供有效的技术支持。
Abstract：Efficient prediction of synergy in anticancer drug combinations is key to developing personalized combination therapy regimens. The biological activity of anticancer drugs relies highly on pharmacophores within their molecular structures. However, existing computational models have limitations in deeply mining and utilizing pharmacophore features within molecular structures. To address this issue, this study proposes a multi-scale molecular structural similarity-based synergy prediction model for anticancer drug combinations( MSSynergy). The model utilizes a three-layer graph convolutional network to hierarchically characterize the multi-scale structural features of drug molecules, introduces an attention mechanism to quantify structural similarity scores of drug pairs at different scales to capture key antitumor efficacy features from the perspective of structural similarity, utilizes multi-layer convolutional neural networks to extract transcriptomic gene expression features of cancer cell lines, and achieves end-to-end prediction of synergy scores or classifications via a fully connected layer. Experimental results show that the mean squared error( MSE), Spearman correlation coefficient( SCC), coefficient of determination(R2), area under the receiver operating characteristic curve( AUC), area under the precision-recall curve( AUPR), and accuracy( ACC) of MSSynergy reached 0. 010, 0. 836, 0. 994, 0. 940, 0. 939, and 0. 870, respectively, all superior to existing methods, fully validating that the model MSSynergy can provide effective technical support for precision cancer therapy and clinical drug screening.
Barretina J,Caponigro G,Stransky N,et al.,2012.The cancer cell line encyclopedia enables predictive modelling of anticancer drug sensitivity[J].Nature,483(7391):603-607.
Bento A P,Hersey A,Félix E,et al.,2020.An open source chemical structure curation pipeline using RDKit[J].J.Cheminform.,12(1):51.
Celebi R,Bear Don′t Walk O4th,Movva R,et al.,2019.Insilico prediction of synergistic anti-cancer drug combinations using multi-omics data[J].Sci.Rep.,9(1):8949.
Han K,Cao P G,Wang Y,et al.,2022.A review of approaches for predicting drug-drug interactions based on machine learning[J].Front.Pharmacol.,12:814858.
He L Y,Kulesskiy E,Saarela J,et al.,2018.Methods for highthroughput drug combination screening and synergy scoring[M].New York:Humana Press:351-398.
Ianevski A,Giri A K,Gautam P,et al.,2019.Prediction of drug combination effects with a minimal set of experiments[J].Nat.Mach.Intell.,1(12):568-577.
Kim S,Chen J,Cheng T J,et al.,2019.Pub Chem 2019 update:improved access to chemical data[J].Nucleic Acids Res.,47(D1):D1102-D1109.
Kipf T N,Welling M,2016.Semi-supervised classification with graph convolutional networks[PP/OL].ar Xiv (2016-09-09)[2026-03-04].https://arxiv.org/abs/1609.02907.
Kolesnikov N,Hastings E,Keays M,et al.,2015.Array Express update:simplifying data submissions[J].Nucleic Acids Res.,43(D1):D1113-D1116.
Kuru H I,Tastan O,Cicek A E,2022.MatchMaker:a deep learning framework for drug synergy prediction[J].IEEE/ACM Trans.Comput.Biol.Bioinform.,19(4):2334-2344.
Li L,Li H T,Ishdorj T O,et al.,2024.MDNNSyn:a multi-modal deep learning framework for drug synergy prediction[J].IEEE J.Biomed.Health Inform.,28(10):6225-6236.
Liu X,Song C Z,Liu S C,et al.,2022.Multi-way relationenhanced hypergraph representation learning for anti-cancer drug synergy prediction[J].Bioinformatics,38(20):4782-4789.
Liu Y J,Zhang P L,Che C,et al.,2024.SDDSynergy:learning important molecular substructures for explainable anticancer drug synergy prediction[J].J.Chem.Inf.Model.,64(24):9551-9562.
Martin Y C,Kofron J L,Traphagen L M,2002.Do structurally similar molecules have similar biological activity?[J].J.Med.Chem.,45(19):4350-4358.
Menden M P,Wang D,Mason M J,et al.,2019.Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen[J].Nat.Commun.,10:2674.
O′Neil J,Benita Y,Feldman I,et al.,2016.An unbiased oncology compound screen to identify novel combination strategies[J].Mol.Cancer Ther.,15(6):1155-1162.
Presti D,Quaquarini E,2019.The PI3K/AKT/m TOR and CDK4/6 pathways in endocrine resistant HR+/HER2-metastatic breast cancer:biological mechanisms and new treatments[J].Cancers,11(9):1242.
Preuer K,Lewis R P I,Hochreiter S,et al.,2018.Deep Synergy:predicting anti-cancer drug synergy with Deep Learning[J].Bioinformatics,34(9):1538-1546.
Rafiei F,Zeraati H,Abbasi K,et al.,2023.Deep Tra Synergy:drug combinations using multimodal deep learning with transformers[J].Bioinformatics,39(8):btad438.
Rani P,Dutta K,Kumar V,2022.Artificial intelligence techniques for prediction of drug synergy in malignant diseases:past,present,and future[J].Comput.Biol.Med.,144:105334.
Seo H,Tkachuk D,Ho C,et al.,2020.SYNERGx DB:an integrative pharmacogenomic portal to identify synergistic drug combinations for precision oncology[J].Nucleic Acids Res.,48(W1):W494-W501.
Subramanian A,Narayan R,Corsello S M,et al.,2017.A next generation connectivity map:L1000 platform and the first1,000,000 profiles[J].Cell,171(6):1437-1452.e17.
Wildenhain J,Spitzer M,Dolma S,et al.,2015.Prediction of synergism from chemical-genetic interactions by machine learning[J].Cell Syst.,1(6):383-395.
Zhang Q Q,Zhang S W,Feng Y H,et al.,2023.Few-shot drug synergy prediction with a prior-guided hypernetwork architecture[J].IEEE Trans.Pattern Anal.Mach.Intell.,45(8):9709-9725.
Zhang T Y,Zhang L W,Payne P R O,et al.,2021.Synergistic drug combination prediction by integrating multiomics data in deep learning models[M]//Translational Bioinformatics for Therapeutic Development.New York,NY:Springer US:223-238.
Zhao S,Nishimura T,Chen Y B,et al.,2013.Systems pharmacology of adverse event mitigation by drug combinations[J].Sci.Transl.Med.,5(206):206ra140.
基本信息:
DOI：10.13417/j.gab.045.000801
中图分类号:R979.1
引用信息:
[1]郭林扬,黄毅然,张锦雄,等.基于药物多尺度分子结构相似性的抗癌药物组合协同性预测[J].基因组学与应用生物学,2026,45(04):801-809.DOI:10.13417/j.gab.045.000801.
基金信息:
国家自然科学基金项目(62362004,61862006); 广西自然科学基金项目(2025GXNSFAA069107,2020GXNSFAA159074)共同资助
2026-05-26
2026-05-26
2026-05-26