Abstract
Complex diseases, such as cancer and neurodegeneration, feature interconnected pathways, making single-target therapies ineffective due to pathway redundancy and compensatory mechanisms. Polypharmacy, which combines multiple drugs to target distinct proteins, addresses this but often leads to drug-drug interactions, cumulative toxicity, and complex pharmacokinetics. To overcome these challenges, we introduce EVOSYNTH, a modular framework for multi-target drug discovery that combines latent evolution and synthesis-aware prioritization to generate and prioritize candidates with high translational potential. Latent evolution navigates a chemically and functionally informed latent space to identify candidates with strong predicted affinity across multiple targets. Synthesis-aware prioritization evaluates both retrosynthetic feasibility and the trade-off between synthetic cost and therapeutic reward, enabling practical and efficient candidate selection. Applied to dual inhibition of JNK3 and GSK3β in Alzheimer’s disease and PI3K and PARP1 in ovarian cancer, EVOSYNTH consistently outperforms baseline generative models, achieving higher predicted affinities, improved scaffold diversity, and lower synthesis costs. These findings highlight EVOSYNTH’s ability to integrate target-driven generation with practical synthesizability, establishing a scalable framework for multi-target and polypharmacological drug discovery. Our source code and data to reproduce all experiments are publicly available on GitHub at: https://github.com/HySonLab/EvoSynth.
Similar content being viewed by others
Data availability
All datasets used in this study are derived from publicly accessible sources. Processed data and docking results necessary to reproduce the findings are publicly available at https://github.com/HySonLab/EvoSynth.
Code availability
The source code for EVOSYNTH, including inference pipelines, and retrosynthesis analysis tools, is publicly available at https://github.com/HySonLab/EvoSynth. Pretrained EVOSYNTH checkpoints required for inference are publicly available on Zenodo at https://zenodo.org/record/17351094.
References
Ramsay, R. R., Popovic-Nikolic, M. R., Nikolic, K., Uliassi, E. & Bolognesi, M. L. A perspective on multi-target drug discovery and design for complex diseases. Clin. Transl. Med. 7, 3 (2018).
Halli-Tierney, A. D., Scarbrough, C. & Carroll, D. Polypharmacy: evaluating risks and deprescribing. Am. Fam. Physician 100, 32–38 (2019).
Makhoba, X. H., Viegas Jr, C., Mosa, R. A., Viegas, F. P. & Pooe, O. J. Potential impact of the multi-target drug approach in the treatment of some complex diseases. Drug Des. Develop. Ther. 14, 3235–3249 (2020).
Katsoulaki, E.-E., Dimopoulos, D. & Hadjipavlou-Litina, D. Multitarget compounds designed for alzheimer, parkinson, and huntington neurodegeneration diseases. Pharmaceuticals 18, 831 (2025).
Doostmohammadi, A., Jooya, H., Ghorbanian, K., Gohari, S. & Dadashpour, M. Potentials and future perspectives of multi-target drugs in cancer treatment: the next generation anti-cancer agents. Cell Commun. Signal. 22, 228 (2024).
Zhang, K. et al. Artificial intelligence in drug development. Nat. Med. 31, 45–59 (2025).
Tanoli, Z. et al. Computational drug repurposing: approaches, evaluation of in silico resources and case studies. Nat. Rev. Drug Discov. 24, 521–542 (2025).
Savander, S. et al. Data-driven strategies for drug repurposing: insights, recommendations, and case studies. Brief. Bioinforma. 26, bbaf625 (2025).
Reinshagen, J. et al. From library to landscape: integrative annotation workflows for compound libraries in drug repurposing. Database 2025, baaf081 (2025).
Ianevski, A. et al. Repurposedrugs: an interactive web-portal and predictive platform for repurposing mono-and combination therapies. Brief. Bioinforma. 25, bbae328 (2024).
Aldahdooh, J., Tanoli, Z. & Tang, J. Mining drug–target interactions from biomedical literature using chemical and gene descriptions-based ensemble transformer model. Bioinforma. Adv. 4, vbae106 (2024).
Aldahdooh, J., Vähä-Koskela, M., Tang, J. & Tanoli, Z. Using bert to identify drug-target interactions from whole pubmed. BMC Bioinforma. 23, 245 (2022).
Legehar, A. et al. Ecodrug plus: an advanced database for drug target conservation analysis and environmental risk assessment. Nucleic Acids Res. 54, D1397–D1404 (2026).
Tang, X. et al. A survey of generative AI for de novo drug design: new frontiers in molecule and protein generation. Brief. Bioinf. 25, bbae338 (2024).
Kang, S. I., Shin, J. H., Wu, B. M. & Choi, H. S. Deep generative AI for multi-target therapeutic design: Toward self-improving drug discovery framework. Int. J. Mol. Sci. 26, 11443 (2025).
Zou, Y. et al. A structure-based framework for selective inhibitor design and optimization. Commun. Biol. 8, 422 (2025).
Stiller, T., Duensing-Kropp, S., Marschner, J. A. & Merk, D. Fragment-based discovery of dual ligand pharmacophores for lipid-sensing transcription factors for designed polypharmacology. RSC Med. Chem. 16, 5012–5024 (2025).
Isigkeit, L. et al. Automated design of multi-target ligands by generative deep learning. Nat. Commun. 15, 7946 (2024).
Liu, X. et al. Drugex v2: de novo design of drug molecules by pareto-based multi-objective reinforcement learning in polypharmacology. J. Cheminf. 13, 85 (2021).
Li, Y., Lin, X., Hao, Y., Zhang, J. & Gao, Y. Q. Molsculptor: a diffusion-evolution framework for multi-site inhibitor design. ChemRxiv (2025).
Ertl, P. & Schuffenhauer, A. Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity and fragment contributions. J. Cheminf. 1, 8 (2009).
Bickerton, G. R., Paolini, G. V., Besnard, J., Muresan, S. & Hopkins, A. L. Quantifying the chemical beauty of drugs. Nat. Chem. 4, 90–98 (2012).
Swanson, K. et al. Generative ai for designing and validating easily synthesizable and structurally novel antibiotics. Nat. Mach. Intell. 6, 338–353 (2024).
Koziarski, M. et al. Rgfn: Synthesizable molecular generation using gflownets. Adv. Neural Inf. Process. Syst. 37, 46908–46955 (2024).
Coley, C. W., Barzilay, R., Jaakkola, T. S., Green, W. H. & Jensen, K. F. Prediction of organic reaction outcomes using machine learning. ACS Cent. Sci. 3, 434–443 (2017).
Genheden, S. et al. Aizynthfinder: a fast, robust and flexible open-source software for retrosynthetic planning. J. Cheminf. 12, 70 (2020).
Fromer, J. C. & Coley, C. W. An algorithmic framework for synthetic cost-aware decision making in molecular design. Nat. Comput. Sci. 4, 440–450 (2024).
Kim, H. et al. Targeting jnk3 for alzheimer’s disease: Design and synthesis of novel inhibitors with aryl group diversity utilizing wide pocket. Eur. J. Med. Chem. 285, 117209 (2025).
Wang, D. et al. Combined inhibition of pi3k and parp is effective in the treatment of ovarian cancer cells with wild-type PIK3CA genes. Gynecol. Oncol. 142, 548–556 (2016).
Scheltens, P. et al. Alzheimer’s disease. Lancet 397, 1577–1590 (2021).
Gong, C.-X., Liu, F. & Iqbal, K. Multifactorial hypothesis and multi-targets for Alzheimer’s disease. J. Alzheimer’s. Dis. 64, S107–S117 (2018).
Probst, G. D. et al. Highly selective c-jun n-terminal kinase (jnk) 2 and 3 inhibitors with in vitro cns-like pharmacokinetic properties prevent neurodegeneration. Bioorg. Med. Chem. Lett. 21, 315–319 (2011).
Buonfiglio, R. et al. Discovery of novel imidazopyridine gsk-3β inhibitors supported by computational approaches. Molecules 25, 2163 (2020).
Shuai, W. et al. Discovery of novel indazole chemotypes as isoform-selective jnk3 inhibitors for the treatment of parkinson’s disease. J. Med. Chem. 66, 1273–1300 (2023).
Jayson, G. C., Kohn, E. C., Kitchener, H. C. & Ledermann, J. A. Ovarian cancer. Lancet 384, 1376–1388 (2014).
Murawski, M., Jagodziński, A., Bielawska-Pohl, A. & Klimczak, A. Complexity of the genetic background of oncogenesis in ovarian cancer-genetic instability and clinical implications. Cells 13, 345 (2024).
Walker, E. H. et al. Structural determinants of phosphoinositide 3-kinase inhibition by wortmannin, ly294002, quercetin, myricetin, and staurosporine. Mol. Cell 6, 909–919 (2000).
Ogden, T. E. et al. Dynamics of the hd regulatory subdomain of parp-1; substrate access and allostery in parp activation and inhibition. Nucleic Acids Res. 49, 2266–2288 (2021).
Oord, A. v. d., Li, Y. & Vinyals, O. Representation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748 (2018).
Castro, E. et al. Transformer-based protein generation with regularized latent space optimization. Nat. Mach. Intell. 4, 840–851 (2022).
Schulman, A., Rousu, J., Aittokallio, T. & Tanoli, Z. Attention-based approach to predict drug–target interactions across seven target superfamilies. Bioinformatics 40, btae496 (2024).
McNutt, A. T. et al. Sprint enables interpretable and ultra-fast virtual screening against thousands of proteomes. arXiv e-prints arXiv–2411 (2024).
Passaro, S. et al. Boltz-2: Towards accurate and efficient binding affinity prediction. BioRxiv 2025–06 (2025).
Chen, T., Kornblith, S., Norouzi, M. & Hinton, G. A simple framework for contrastive learning of visual representations. In Proc. International Conference on Machine Learning, 1597–1607 (PMLR, 2020).
Huang, Y. et al. Dsdp: A blind docking strategy accelerated by gpus. J. Chem. Inf. Model. 63, 4355–4363 (2023).
RDKit — rdkit.org. https://www.rdkit.org/.
Deb, K., Pratap, A., Agarwal, S. & Meyarivan, T. A fast and elitist multiobjective genetic algorithm: Nsga-ii. IEEE Trans. Evolut. Comput. 6, 182–197 (2002).
Chemspace services: compound sourcing and procurement, hit discovery, molecular docking, custom synt—chem-space.com. https://chem-space.com/services.
Author information
Authors and Affiliations
Contributions
T.S.H. and V.T.D.N. conceived the research idea. V.T.D.N. developed the EVOSYNTH framework, implemented the experiments, and performed data analysis. P.P. contributed to conceptual refinement, evaluation design and code optimization. V.T.D.N. and P.P. jointly wrote the initial manuscript draft. T.S.H. supervised the project and provided critical feedback and revisions. All authors discussed the results and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
: Communications Chemistry thanks Qiaoyu Hu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Nguyen, V.T.D., Pham, P. & Hy, TS. Enabling multi-target drug discovery through latent evolutionary optimization and synthesis-aware prioritization (EVOSYNTH). Commun Chem (2026). https://doi.org/10.1038/s42004-026-01945-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s42004-026-01945-4
