Academic Journal of Computing & Information Science, 2026, 9(3); doi: 10.25236/AJCIS.2026.090309.
Qihao Wang1, Yanxiang Xie1, Bin Song2
1School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai, 200093, China
2Department of Hepatobiliary, Pancreatic, and Spleen Surgery, Changhai Hospital, Naval Medical University, Shanghai, 200433, China
This study aims to screen Chinese herbal medicine monomers with anti-pancreatic cancer activity and establish a compound screening model using graph neural network deep learning algorithms to provide a novel strategy for large-scale screening of anti-pancreatic cancer herbal monomers, first employing CTG technology to detect the proliferation-inhibitory effects of 970 TCM monomers on BxPC-3 pancreatic cancer cells, then taking this data as training material to construct an anti-pancreatic cancer activity prediction model within the Chemprop framework, applying this model to predict anti-pancreatic cancer activity for over 30,000 natural product molecules in the TCMBank database, and conducting experimental validation on the top five TCM monomers with the highest predicted scores; the results showed that 87 of the 970 TCM monomers (approximately 9.0% of the total) exhibited >80% inhibition of BxPC-3 cell proliferation, the constructed model achieved an R2 of 0.81 and an RMSE of 11.34 on the test set with excellent performance, and three of the five candidate compounds (Alkannin, Rottlerin, and β-Mangostin) selected for experimental validation exhibited significant, dose-dependent inhibitory effects on BxPC-3 cells; this study evaluated the anti-pancreatic cancer activity of 970 TCM monomers and used the results as a training set to construct a deep learning model, which enabled large-scale screening of TCM monomers for anti-pancreatic cancer activity and provided a novel strategy for AI-driven drug discovery.
Traditional Chinese medicine; Pancreatic ductal adenocarcinoma; Graph neural networks; Deep learning; Anti-Cancer drug screens
Qihao Wang, Yanxiang Xie, Bin Song. Construction of a Graph Neural Network-Based Activity Prediction Model for Traditional Chinese Medicine Monomers Against Pancreatic Cancer. Academic Journal of Computing & Information Science (2026), Vol. 9, Issue 3: 72-81. https://doi.org/10.25236/AJCIS.2026.090309.
[1] SIEGEL R L, GIAQUINTO A N, JEMAL A. Cancer statistics, 2024[J/OL]. CA: A Cancer Journal for Clinicians, 2024, 74(1): 12-49. DOI:10.3322/caac.21820.
[2] CONNOR A A, GALLINGER S. Pancreatic cancer evolution and heterogeneity: integrating omics and clinical data[J/OL]. Nature Reviews Cancer, 2022, 22(3): 131-142. DOI:10.1038/s41568-021-00418-1.
[3] ZHAO W, ZHENG X D, TANG P Y Z, et al. Advances of antitumor drug discovery in traditional Chinese medicine and natural active products by using multi‐active components combination [J/OL]. [2025] [2025-08-25]. https://onlinelibrary.wiley.com/doi/10.1002/med.21963.DOI:10.1002/med.21963.
[4] AN Q, HUANG L, WANG C, et al. New strategies to enhance the efficiency and precision of drug discovery [J/OL]. Frontiers in Pharmacology, 2025, 16[2025-08-25]. https://www.frontiersin.org/journals/ pharmacology/articles/10.3389/fphar.2025.1550158/full. DOI:10.3389/fphar.2025.1550158.
[5] VAMATHEVAN J, CLARK D, CZODROWSKI P, et al. Applications of machine learning in drug discovery and development[J/OL]. Nature Reviews Drug Discovery, 2019, 18(6): 463-477. DOI:10.1038/s41573-019-0024-5.
[6] YANG K, SWANSON K, JIN W, et al. Analyzing Learned Molecular Representations for Property Prediction[J/OL]. Journal of Chemical Information and Modeling, 2019[2025-08-25]. https://pubs.acs.org/ doi/full/10.1021/acs.jcim.9b00237. DOI:10.1021/acs.jcim.9b00237.
[7] WEININGER D, WEININGER A. SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules [J]. Journal of Chemical Information and Computer Sciences, 1988, 28 (1): 31-36. DOI:10.1021/ci00057a005.
[8] HEID E, GREENMAN K P, CHUNG Y, et al. Chemprop: A Machine Learning Package for Chemical Property Prediction[J/OL]. Journal of Chemical Information and Modeling, 2023 [2025-08-25]. https://pubs.acs.org/doi/full/10.1021/acs.jcim.3c01250. DOI:10.1021/acs.jcim.3c01250.
[9] WONG F, OMORI S, DONGHIA N M, et al. Discovering small-molecule senolytics with deep neural networks[J/OL]. Nature Aging, 2023, 3(6): 734-750. DOI:10.1038/s43587-023-00415-z.
[10] NOTWELL J H, WOOD M W. ADMET property prediction through combinations of molecular fingerprints[A/OL]. arXiv, 2023[2025-08-25]. http://arxiv.org/abs/2310.00174. DOI:10.48550/arXiv. 2310.00174.
[11] ISERT C, KROMANN J C, STIEFL N, et al. Machine Learning for Fast, Quantum Mechanics-Based Approximation of Drug Lipophilicity[J/OL]. ACS Omega, 2023[2025-08-25]. https://pubs.acs.org/doi/full/ 10.1021/acsomega.2c05607. DOI:10.1021/acsomega.2c05607.
[12] LV Q, CHEN G, HE H, et al. TCMBank-the largest TCM database provides deep learning-based Chinese-Western medicine exclusion prediction[J/OL]. Signal Transduction and Targeted Therapy, 2023, 8(1): 127. DOI:10.1038/s41392-023-01339-1.
[13] OMURA S, SASAKI Y, IWAI Y, et al. Staurosporine, a Potentially Important Gift from a Microorganism[J/OL]. The Journal of Antibiotics, 1995, 48(7): 535-548. DOI:10.7164/antibiotics.48.535.
[14] ZHANG Y, SUI X, PAN F, et al. A comprehensive large-scale biomedical knowledge graph for AI-powered data-driven biomedical research[J/OL]. Nature Machine Intelligence, 2025, 7(4): 602-614. DOI:10.1038/s42256-025-01014-w.
[15] FANG J, LIU D, TSE C K. Impact of Structure of Network Based Data on Performance of Graph Neural Networks[C/OL]//2023 IEEE International Symposium on Circuits and Systems (ISCAS). 2023: 1-5 [2025-08-25]. https://ieeexplore.ieee.org/abstract/document/10182188. DOI:10.1109/ISCAS46773.2023.10182188.
[16] EDWARDS W, BARRON F H. SMARTS and SMARTER: Improved Simple Methods for Multiattribute Utility Measurement[J/OL]. Organizational Behavior and Human Decision Processes, 1994, 60(3): 306-325. DOI:10.1006/obhd.1994.1087.
[17] PROBST D, REYMOND J L. A probabilistic molecular fingerprint for big data settings[J/OL]. Journal of Cheminformatics, 2018, 10(1): 66. DOI:10.1186/s13321-018-0321-8.
[18] MCINNES L, HEALY J, MELVILLE J. UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction[A/OL]. arXiv, 2020[2025-08-25]. http://arxiv.org/abs/1802.03426. DOI:10.48550/arXiv.1802.03426.
[19] CAI L, QIN X, XU Z, et al. Comparison of Cytotoxicity Evaluation of Anticancer Drugs between Real-Time Cell Analysis and CCK-8 Method[J/OL]. ACS Omega, 2019[2025-08-25]. https://pubs.acs.org/doi/full/10.1021/acsomega.9b01142. DOI:10.1021/acsomega.9b01142.