A multimodal contrastive learning framework for predicting P-glycoprotein substrates and inhibitors

Yixue Zhang; Jialu Wu; Yu Kang; Tingjun Hou

doi:10.1016/j.jpha.2025.101313

Volume 15 Issue 8

Sep. 2025

Turn off MathJax

Article Contents

Article Navigation > Journal of Pharmaceutical Analysis > 2025 > 15(8): 101313

Yixue Zhang, Jialu Wu, Yu Kang, Tingjun Hou. A multimodal contrastive learning framework for predicting P-glycoprotein substrates and inhibitors[J]. Journal of Pharmaceutical Analysis, 2025, 15(8): 101313. doi: 10.1016/j.jpha.2025.101313

Citation:

Yixue Zhang, Jialu Wu, Yu Kang, Tingjun Hou. A multimodal contrastive learning framework for predicting P-glycoprotein substrates and inhibitors[J]. Journal of Pharmaceutical Analysis, 2025, 15(8): 101313. doi: 10.1016/j.jpha.2025.101313

Citation:

PDF( 3731 KB)

A multimodal contrastive learning framework for predicting P-glycoprotein substrates and inhibitors

doi: 10.1016/j.jpha.2025.101313

Yixue Zhang^a,b,
Jialu Wu^a,
Yu Kang^{a
,
,},
Tingjun Hou^{a
,
,}

a. College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, 310058, China;
b. Polytechnic Institute of Zhejiang University, Zhejiang University, Hangzhou, 310015, China

Funds:

This work was financially supported by the National Key Research and Development Program of China (Program No.: 2022YFF1203003), and the National Natural Science Foundation of China (Grant No.: 82373791).

Received Date: Sep. 24, 2024
Accepted Date: Apr. 10, 2025
Rev Recd Date: Mar. 07, 2025
Publish Date: Apr. 16, 2025

Abstract

Abstract

P-glycoprotein (P-gp) is a transmembrane protein widely involved in the absorption, distribution, metabolism, excretion, and toxicity (ADMET) of drugs within the human body. Accurate prediction of P-gp inhibitors and substrates is crucial for drug discovery and toxicological assessment. However, existing models rely on limited molecular information, leading to suboptimal model performance for predicting P-gp inhibitors and substrates. To overcome this challenge, we compiled an extensive dataset from public databases and literature, consisting of 5,943 P-gp inhibitors and 4,018 substrates, notable for their high quantity, quality, and structural uniqueness. In addition, we curated two external test sets to validate the model's generalization capability. Subsequently, we developed a multimodal graph contrastive learning (GCL) model for the prediction of P-gp inhibitors and substrates (MC-PGP). This framework integrates three types of features from Simplified Molecular Input Line Entry System (SMILES) sequences, molecular fingerprints, and molecular graphs using an attention-based fusion strategy to generate a unified molecular representation. Furthermore, we employed a GCL approach to enhance structural representations by aligning local and global structures. Extensive experimental results highlight the superior performance of MC-PGP, which achieves improvements in the area under the curve of receiver operating characteristic (AUC-ROC) of 9.82% and 10.62% on the external P-gp inhibitor and external P-gp substrate datasets, respectively, compared with 12 state-of-the-art methods. Furthermore, the interpretability analysis of all three molecular feature types offers comprehensive and complementary insights, demonstrating that MC-PGP effectively identifies key functional groups involved in P-gp interactions. These chemically intuitive insights provide valuable guidance for the design and optimization of drug candidates.
- P-glycoprotein,
- Deep learning,
- Multimodal fusion,
- Graph contrastive learning

FullText(HTML)

References(47)

References

[1]	I.I. Ahmed Juvale, A.A. Abdul Hamid, K.B. Abd Halim, et al., P-glycoprotein: New insights into structure, physiological function, regulation and alterations in disease, Heliyon 8 (2022), e09777.
[2]	M. Elmeliegy, M. Vourvahis, C. Guo, et al., Effect of P-glycoprotein (P-gp) inducers on exposure of P-gp substrates: Review of clinical drug-drug interaction studies, Clin. Pharmacokinet. 59 (2020) 699-714.
[3]	C. Karthika, R. Sureshkumar, M. Zehravi, et al., Multidrug resistance of cancer cells and the vital role of P-glycoprotein, Life (Basel) 12 (2022), 897.
[4]	J. Yabut, R. Houle, S. Wang, et al., Selection of an optimal in vitro model to assess P-gp inhibition: Comparison of vesicular and bidirectional transcellular transport inhibition assays, Drug Metab. Dispos. 50 (2022) 909-922.
[5]	Z. Liu, K. Liu, The transporters of intestinal tract and techniques applied to evaluate interactions between drugs and transporters, Asian J. Pharm. Sci. 8 (2013) 151-158.
[6]	T.T.V. Tran, H. Tayara, K.T. Chong, Recent studies of artificial intelligence on in silico drug absorption, J. Chem. Inf. Model. 63 (2023) 6198-6211.
[7]	C. Esposito, S. Wang, U.E.W. Lange, et al., Combining machine learning and molecular dynamics to predict P-glycoprotein substrates, J. Chem. Inf. Model. 60 (2020) 4730-4749.
[8]	D.S. Wigh, J.M. Goodman, A.A. Lapkin, A review of molecular representation in the age of machine learning, Wiley Interdiscip. Rev. Comput. Mol. Sci. 12 (2022), e1603.
[9]	Z. Li, M. Jiang, S. Wang, et al., Deep learning methods for molecular representation and property prediction, Drug Discov. Today 27 (2022), 103373.
[10]	A.L. Nazarova, L. Yang, K. Liu, et al., Dielectric polymer property prediction using recurrent neural networks with optimizations, J. Chem. Inf. Model. 61 (2021) 2175-2186.
[11]	Z. Wu, D. Jiang, J. Wang, et al., Knowledge-based BERT: A method to extract molecular features like computational chemists, Brief. Bioinform. 23 (2022), bbac131.
[12]	J. Ross, B. Belgodere, V. Chenthamarakshan, et al., Large-scale chemical language representations capture molecular structure and properties, Nat. Mach. Intell. 4 (2022) 1256-1264.
[13]	J. Deng, Z. Yang, H. Wang, et al., A systematic study of key elements underlying molecular property prediction, Nat. Commun. 14 (2023), 6395.
[14]	Y. Wei, S. Li, Z. Li, et al., Interpretable-ADMET: A web service for ADMET prediction and optimization based on deep neural representation, Bioinformatics 38 (2022) 2863-2871.
[15]	Z. Xiong, D. Wang, X. Liu, et al., Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism, J. Med. Chem. 63 (2020) 8749-8760.
[16]	Z. Wu, D. Jiang, J. Wang, et al., Mining toxicity information from large amounts of toxicity data, J. Med. Chem. 64 (2021) 6924-6936.
[17]	Y. Jiang, S. Jin, X. Jin, et al., Pharmacophoric-constrained heterogeneous graph transformer model for molecular property prediction, Commun. Chem. 6 (2023), 60.
[18]	B. Zhao, W. Xu, J. Guan, et al., Molecular property prediction based on graph structure learning, Bioinformatics 40 (2024), btae304.
[19]	G.P. Wellawatte, H.A. Gandhi, A. Seshadri, et al., A perspective on explanations of molecular prediction models, J. Chem. Theory Comput. 19 (2023) 2149-2160.
[20]	P.-H. Wang, Y.-S. Tu, Y. J. Tseng, PgpRules: A decision tree based prediction server for P-glycoprotein substrates and inhibitors, Bioinformatics 35 (2019) 4193-4195.
[21]	A. Sedykh, D. Fourches, J. Duan, et al., Human intestinal transporter database: QSAR modeling and virtual profiling of drug uptake, efflux and interactions, Pharm. Res. 30 (2013) 996-1007.
[22]	V. Namasivayam, K. Silbermann, M. Wiese, et al., C@PA: Computer-aided pattern analysis to predict multitarget ABC transporter inhibitors, J. Med. Chem. 64 (2021) 3350-3366.
[23]	F. Broccatelli, E. Carosati, A. Neri, et al., A novel approach for predicting P-glycoprotein (ABCB1) inhibition using molecular interaction fields, J. Med. Chem. 54 (2011) 1740-1751.
[24]	D. Li, L. Chen, Y. Li, et al., ADMET evaluation in drug discovery. 13.Development of in silico prediction models for P-glycoprotein substrates, Mol. Pharm. 11 (2014) 716-726.
[25]	J.W. Polli, S.A. Wring, J.E. Humphreys, et al., Rational use of in vitro P-glycoprotein assays in drug discovery, J. Pharmacol. Exp. Ther. 299 (2001) 620-628.
[26]	L. Mora Lagares, N. Minovski, M. Novic, Multiclass classifier for P-glycoprotein substrates, inhibitors, and non-active compounds, Molecules 24 (2019), 2006.
[27]	G. Xiong, Z. Wu, J. Yi, et al., ADMETlab 2.0: An integrated online platform for accurate and comprehensive predictions of ADMET properties, Nucleic Acids Res. 49 (2021) W5-W14.
[28]	Organisation for Economic Co-operation and Development, Assessment of (Q)SAR Models (Model Checklist). (Q)SAR Assessment Framework: Guidance for the Regulatory Assessment of (Quantitative) Structure-Activity Relationship Models, Predictions, and Results Based on Multiple Predictions, OECD Series on Testing and Assessment, No. 386, OECD Publishing, Paris, 2024, pp. 19-25.
[29]	J. Jo, B. Kwak, H.-S. Choi, et al., The message passing neural networks for chemical property prediction on SMILES, Methods 179 (2020) 65-72.
[30]	G.B. Goh, N.O. Hodas, C. Siegel, et al., SMILES2Vec: An interpretable general-purpose deep neural network for predicting chemical properties, arXiv. 2017. https://arxiv.org/abs/1712.02034.
[31]	K. Xu, W. Hu, J. Leskovec, et al., How powerful are graph neural networks, arXiv. 2018. https://arxiv.org/abs/1810.00826.
[32]	A. Vaswani, N. Shazeer, N. Parmar, et al., Proceedings of the 31st International Conference on Neural Information Processing Systems, Dec. 3-9, 2017, Curran Associates Inc., New York, 2017, pp. 6000-6010.
[33]	Y. Wang, J. Wang, Z. Cao, et al., Molecular contrastive learning of representations via graph neural networks, Nat. Mach. Intell. 4 (2022) 279-287.
[34]	Y. Yin, Q. Wang, S. Huang, et al., AutoGCL: Automated graph contrastive learning via learnable view generators, arXiv. 2022. https://doi.org/10.48550/arXiv.2109.10259.
[35]	T. Chen, S. Kornblith, M. Norouzi, et al., Proceedings of the 37th International Conference on Machine Learning, July 12-18, 2020, JMLR, Cambridge, 2020, pp. 1597-1607.
[36]	K. Yang, K. Swanson, W. Jin, et al., Analyzing learned molecular representations for property prediction, J. Chem. Inf. Model. 59 (2019) 3370-3388.
[37]	R.R. Selvaraju, M. Cogswell, A. Das, et al., Grad-CAM: Visual explanations from deep networks via gradient-based localization, October 22-29, 2017, Venice, Italy, 2017, pp. 618-626.
[38]	M. Sundararajan, A. Taly, Q. Yan, Proceedings of the 34th International Conference on Machine Learning, August 6-11, 2017, Sydney, Australia, 2017, pp. 3319-3328.
[39]	Y. Hu, D. Stumpfe, J. Bajorath, Computational exploration of molecular scaffolds in medicinal chemistry, J. Med. Chem. 59 (2016) 4062-4076.
[40]	R.B. Wang, C.L. Kuo, L.L. Lien, et al., Structure-activity relationship: Analyses of p-glycoprotein substrates and inhibitors, J. Clin. Pharm. Ther. 28 (2003) 203-228.
[41]	H. Cai, H. Zhang, D. Zhao, et al., FP-GNN: A versatile deep learning architecture for enhanced molecular property prediction, Brief. Bioinform. 23 (2022), bbac408.
[42]	W. Zhu, Y. Zhang, D. Zhao, et al., HiGNN: A hierarchical informative graph neural network for molecular property prediction equipped with feature-wise attention, J. Chem. Inf. Model. 63 (2023) 43-55.
[43]	J. Jiang, R. Zhang, Z. Zhao, et al., MultiGran-SMILES: Multi-granularity SMILES learning for molecular property prediction, Bioinformatics 38 (2022) 4573-4580.
[44]	B.-X. Du, Y. Long, X. Li, et al., CMMS-GCL: Cross-modality metabolic stability prediction with graph contrastive learning, Bioinformatics 39 (2023), btad503.
[45]	T. Wang, Z. Li, L. Zhuo, et al., MS-BACL: Enhancing metabolic stability prediction through bond graph augmentation and contrastive learning, Brief. Bioinform. 25 (2024), bbae127.
[46]	X. Zhao, J. Di, D. Luo, et al., Recent developments of P-glycoprotein inhibitors and its structure-activity relationship (SAR) studies, Bioorg. Chem. 143 (2024), 106997.
[47]	H. Zhang, H. Xu, C.R. Ashby Jr., et al., Chemical molecular-based approach to overcome multidrug resistance in cancer by targeting P-glycoprotein (P-gp), Med. Res. Rev. 41 (2021) 525-555.