Adaptation of the REINVENT neural network architecture to generate potential HIV-1 entry inhibitors

Objectives. The main purpose of this work is to adapt the architecture of the REINVENT neural network to generate potential inhibitors of the HIV-1 envelope protein gp120 using in the learning process with reinforcement of molecular docking on GPUs.

Methods. To modify the initial network model, molecular docking on GPUs implemented in the learning process with reinforcement was used, and an algorithm was developed that allows converting the representations of connections generated by the SMILES network into the PDBQT format necessary for docking. To accelerate the learning of the neural network in the modified version of the REINVENT model, the AutoDock-Vina-GPU-2.1 docking program was used, and to clarify the results of its work, the procedure for revaluing the affinity of compounds to the target using the RFScore-4 evaluation function was used.

Results. Using a modified version of the REINVENT model, more than 60,000 compounds were obtained, of which about 52,000 molecules have a binding energy value to the HIV-1 gp120 protein comparable to the value calculated for the HIV-1 inhibitor NBD-14204, used in calculations as a positive control. Of the 52,000 compounds selected, about 34,000 molecules satisfy the restrictions imposed on a potential drug to ensure its bioavailability when taken orally.

Conclusion. The results obtained allow us to demonstrate the effectiveness of an adapted neural network by the example of designing new potential inhibitors of the gp120 HIV-1 protein capable of blocking the CD4- binding site of the gp120 virus envelope protein and preventing its penetration into host cells.

1. Li H., Sze K.H., Lu ​​G., Ballester P. Machine-learning scoring functions for structure-based virtual screening. Wiley interdisciplinary reviews: Computational Molecular Science., 2020, vol. 11. https://doi.org/10.1002/wcms.1478

2. Xiong G.L., Ye W.L., Shen C., Lu A.P., Hou T.J., Cao D.S. Improving structure-based virtual screening performance via learning from scoring function components. Briefings in Bioinformatics, 2020. https://doi.org/10.1093/bib/bbaa094

3. Stokes J.M., Yang K., Swanson K., Jin W., Cubillos-Ruiz A. A deep learning approach to antibiotic discovery. Cell, 2020, vol. 180, pp. 688−702. https://doi.org/10.1016/j.cell.2020.01.021

4. Timmons P.B., Hewage C. M. ENNAVIA is a novel method which employs neural networks for antiviral and anti-coronavirus activity prediction for therapeutic peptides. Briefings in Bioinformatics, 2021. https://doi.org/10.1093/bib/bbab258

5. Andrianov A.M., Nikolaev G.I., Shuldov N.A., Bosko I.P., Anischenko A.I., Tuzikov A.V. Application of deep learning and molecular modeling to identify small drug-like compounds as potential HIV-1 entry inhibitors. Journal of Biomolecular Structure and Dynamics, 2022, vol. 40, pp. 7555–7573. https://doi.org/10.1080/07391102.2021.1905559

6. Zhang Y., Ye ​​T., Xi H., Juhas M., Li J. Deep learning driven drug discovery: Tackling Severe Acute Respiratory Syndrome Coronavirus 2. Frontiers in Microbiology, 2021. https://doi.org/10.3389/fmicb.2021.739684

7. Kinnings S.L., Liu N., Tonge P.J., Jackson R.M., Xie L., Bourne P.E. A Machine learning-based method to improve docking scoring functions and its application to drug repurposing. Journal of Chemical Information and Modeling, 2011, vol. 51, no. 5, pp. 1195–1197. https://doi.org/10.1021/ci2001346

8. Agastheeswaramoorthy K., Sevilimedu A. Drug REpurposing using AI/ML tools − for Rare Diseases (DREAM-RD): A case study with Fragile X Syndrome (FXS). bioRxiv 2020. https://doi.org/10.1101/2020.09.25.311142

9. Heinzelmann G., Gilson M.K. Automation of absolute protein-ligand binding free energy calculations for docking refinement and compound evaluation. Scientific Reports, 2021, vol. 11, no. 1, pp. 1116. https://doi.org/10.1038/s41598-020-80769-1

10. Meli R., Morris G.M., Biggin P.C. Scoring functions for protein-ligand binding affinity prediction using structure-based deep learning: A review. Frontiers in Bioinformatics, 2022, vol. 2. https://doi.org/10.3389/fbinf.2022.885983

11. Derek P. Metcalf, Glick Z.L., Bortolato A., Jiang A., Cheney D.L., Sherrill C.D. Directional ΔG neural network (Dr ΔG-Net): A modular neural network approach to binding free energy prediction. Journal of Chemical Information and Modeling, 2024, vol. 64, no. 6, pp. 1907-1918. https://doi.org/10.1021/acs.jcim.3c02054

12. Li Y., Fan Z., Rao J., Chen Z., Chu Q., …, Li X. An overview of recent advances and challenges in predicting compound-protein interaction (CPI). Medical Review, 2023, vol. 3, no. 6, pp. 465–486. https://doi.org/10.1515/mr-2023-0030

13. Zhavoronkov A., Ivanenkov Y.A., Aliper A., Veselov M.S., Aladinskiy V.A., …, Aspuru-Guzik A. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nature Biotechnology, 2019, vol. 37, pp. 1038–1040. https://doi.org/10.1038/s41587-019-0224-x

14. Olivecrona M., Blaschke T., Engkvist O., Chen H. J. Molecular de-novo design through deep reinforcement learning. Journal of Cheminformatics, 2017, vol. 9, p. 48. https://doi.org/10.1186/s13321-017-0235-x

15. Blaschke T., Arús-Pous J., Chen H., Margreitter C., Tyrchan C., …, Patronov A. Reinvent 2.0: an AI tool for de novo drug design. Journal of Chemical Information and Modeling, 2020, vol. 60, no. 12, pp. 5918–5922. https://doi.org/10.1021/acs.jcim.0c00915

16. Trott O., Olson A.J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. Journal of Computational Chemistry, 2010, vol. 31, no. 2, pp. 455–461. https://doi.org/10.1002/jcc.21334

17. Tang S., Ding J., Zhu X., Wang Z., Zhao H., Wu J. Vina-GPU 2.1: towards further optimizing docking speed and precision of AutoDock Vina and its derivatives. bioRxiv, 2023. https://doi.org/10.1101/2023.11.04.565429

18. Svensson H., Tyrchan C., Engkvist O., Chehreghani M.H. Utilizing Reinforcement learning for de novo drug design. arXiv, 2023. https://doi.org/10.48550/arXiv.2303.17615

19. Mnih V., Badia A.P., Mirza M., Graves A., Lillicrap T.P., …, Kavukcuoglu K. Asynchronous methods for deep reinforcement learning. Proceedings of The 33rd International Conference on Machine Learning, 2016, vol. 48, pp. 1928–1937. https://proceedings.mlr.press/v48/mniha16.html

20. Weininger D. SMILES, a chemical language and information system. Journal of Chemical Information and Computer Sciences, 1998, vol. 28, pp. 31–36. https://doi.org/10.1021/ci00057a005

21. Berman H.M., Battistuz T., Bhat T.N., Bluhm W.F., Bourne P., …, Zardecki C. The protein data bank. Acta Crystallographica Section D: Biological Crystallography, 2002, vol. 58, no. 6, pp. 899–907. https://doi.org/10.1093/nar/28.1.235

22. Curreli F., Kwon Y.D., Nicolau I., Burgos G., Altieri A., …, Debnath A.K. Antiviral activity and crystal structures of HIV-1 gp120 antagonists. International Journal of Molecular Sciences, 2022, vol. 23, no. 24, p. 15999. https://doi.org/10.3390/ijms232415999

23. Pettersen E.F., Goddard T.D., Huang C.C., Couch G.S., …, Ferrin T.E. UCSF Chimera A visualization system for exploratory research and analysis. Journal of Computational Chemistry, 2004, vol. 25, no. 13, pp. 1605–1612. https://doi.org/10.1002/jcc.20084

24. Kiefer F., Arnold K., Künzli M., Bordoli L., Schwede T. The Swiss-MODEL Repository and associated resources. Nucleic Acids Research, 2009, vol. 37, pp. 387–392. https://doi.org/10.1093/nar/gkn750

25. Benson M.L., Smith R.D., Khazanov N.A., Dimcheff B., Beaver J., …, Carlson H.A. Binding MOAD, a high-quality protein–ligand database. Nucleic Acids Research, 2007, vol. 36, pp. 674-678. https://doi.org/10.1093/nar/gkm911

26. Wójcikowski M., Ballester P.J., Siedlecki P. Performance of machine-learning scoring functions in structure-based virtual screening. Scientific Reports, 2017, vol. 7, no. 1, pp. 1–10. https://doi.org/10.1038/srep46710

27. Lipinski C.A., Lombardo F., Dominy B.W., Feeney P.J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Advanced Drug Delivery Reviews, 2001, vol. 46, no. 1−3, pp. 3–26. https://doi.org/10.1016/S0169-409X(00)00129-0

28. Daina A., Michielin O., Zoete V. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Scientific Reports, 2017, vol. 7. https://doi.org/10.1038/srep42717

29. Veber D.F., Johnson S.R., Cheng H.Y., Smith B.R., Ward K.W., Kopple K.D. Molecular properties that influence the oral bioavailability of drug candidates. Journal of Medicinal Chemistry, 2002, vol. 45, no. 12. pp. 2615-2623. https://doi.org/10.1021/jm020017n