DECODING MOLECULAR PROPERTY SIGNATURES ASSOCIATED WITH CHEMICAL TOXICITY FOR SAFER COMPOUND SCREENING
DOI:
https://doi.org/10.69980/wc92rx52Keywords:
Chemical toxicity, Molecular descriptors, Compound screening, Molecular signatures, Random forestAbstract
Chemical toxicity screening is important in detecting potentially hazardous compounds prior to a detailed experimental evaluation. In this study, molecular property signatures related to chemical toxicity were analyzed, and their use as indicators of safer compound screening was assessed. The molecular descriptors of 171 compounds, divided into toxic and non-toxic, were subjected to statistical comparisons, correlation, principal component analysis and classification models. MDEC-23 exhibited the highest individual correlations with toxicity, while a few structural and physicochemical descriptors provided complementary discriminatory information. Principal component analysis revealed that the variation in the molecular structure was multidimensional with the two first eigenvectors accounting for 45.35% of the total variation. Random forest had higher classification accuracy (0.673) and ROC-AUC (0.659) than logistic regression, but sensitivity to toxic compound was not high enough. The results show that small molecular property signatures are useful as a initial filter of toxicity information for prioritization. This could be useful for more efficient early stage toxicity evaluation and compound selection for safer compounds when adequately validated externally.
References
1.Akermi, S., Dey, A., Li, R., Larzat, N., Idoipe, J. B., Smaoui, S., ... & Sharma, A. (2025). Deep learning applications in genomics-based toxicology assessment. Deep Learning in Genetics and Genomics, 327-344.
2.Badaoui, H., Ouabane, M., Alaqarbeh, M., Elbouhi, M., Lakhlifi, T., & Bouachrine, M. (2025). The main chemical constituents responsible for the antidiabetic properties of the datura metel l plant. decryption and in-silico investigations. Physical Chemistry Research, 13(2), 241-254.
3.Barghi, E. G., Motafeghi, F., Gholami Gharab, J., & Ghassemi Barghi, N. (2026). Decoding DNA damage: artificial intelligence and the future of genotoxicity assessment. All Life, 19(1), 2611210.
4.Bergen, V., Kodella, K., Srikrishnan, S., Barrandon, O., Anderson, S., Rogers-Grazado, M., ... & Zamanighomi, M. (2025). A large-scale human toxicogenomics resource for drug-induced liver injury prediction. Nature Communications, 16(1), 9860.
5.Chen, Y., & Xue, W. (2026). Machine learning for de novo molecular generation: A comprehensive review. ACS chemical neuroscience, 17(4), 666-680.
6.de Sá, A. G., Long, Y., Portelli, S., Pires, D. E., & Ascher, D. B. (2022). toxCSM: comprehensive prediction of small molecule toxicity profiles. Briefings in Bioinformatics, 23(5), bbac337.
7.Encinar, J. A., Yang, C., Fernández-Ginés, R., Tang, Y., & Cuadrado, A. (2026). ADMET-Driven Chemical Navigability Rules for Early-Stage Virtual Compound Prioritization. Journal of Chemical Information and Modeling.
8.Galati, S., Di Stefano, M., Martinelli, E., Macchia, M., Martinelli, A., Poli, G., & Tuccinardi, T. (2022). VenomPred: a machine learning based platform for molecular toxicity predictions. International journal of molecular sciences, 23(4), 2105.
9.Gül, Ş., & Rahim, F. (2021). Toxicity [Data set]. UCI Machine Learning Repository. https://doi.org/10.24432/C59313
10.Kurre, P. N., Muthyala, M. K. K., & Pichika, M. R. (2026). Computational Chemistry and Drug Discovery: Intersection of AI, QSAR, and Molecular Modeling. Computer‐Aided Drug Design: Principles, Techniques, and Applications, 259-280.
11.11.Lejal, V. (2024). Application of artificial intelligence with multi-omics and cell imaging data to decipher chemical-adverse outcomes for human health (Doctoral dissertation, Université Paris Cité).
12.Liu, Y., Zhu, K., Peng, W., Liu, Z., & Mao, X. (2026). Multi-omics and artificial intelligence for precision drug discovery and potential clinical applications. Signal Transduction and Targeted Therapy, 11(1), 210.
13.Liyaqat, T., Ahmad, T., & Saxena, C. (2026). Advancements in Molecular Property Prediction: A Survey of Single and Multimodal Approaches: T. Liyaqat et al. Archives of Computational Methods in Engineering, 33(1), 613-643.
14.Metz, T. O., Adkins, J. N., Armentrout, P. B., Chain, P., Chu, F., Corley, C. D., ... & Wishart, D. (2023). Decoding the Molecular Universe--Workshop Report. arXiv preprint arXiv:2311.11437.
15.Munson, B. P., Chen, M., Bogosian, A., Kreisberg, J. F., Licon, K., Abagyan, R., ... & Ideker, T. (2024). De novo generation of multi-target compounds using deep generative chemistry. Nature Communications, 15(1), 3636.
16.Najaf, A., & Asghar, M. U. (2026). Predicting Efficacy and Toxicity of Phytochemical Drugs with AI Models. In Artificial Intelligence in Phytochemistry: A New Era of Drug Discovery (pp. 83-96). Singapore: Springer Nature Singapore.
17.Ogbodo, J. O., Egba, S. I., Ikechukwu, G. C., Paul, P. C., Mba, J. O., Ugwu, O. P. C., & Ezike, T. C. (2025). Volatile organic compound–drug receptor interactions: a potential tool for drug design in the search for remedies for increasing toxic occupational exposure. Processes, 13(1), 154.
18.Pathan, I., Raza, A., Sahu, A., Joshi, M., Sahu, Y., Patil, Y., & Raza, M. A. (2025). Revolutionizing pharmacology: AI-powered approaches in molecular modeling and ADMET prediction. Medicine in Drug Discovery, 100223.
19.Pires, D. E., Stubbs, K. A., Mylne, J. S., & Ascher, D. B. (2022). cropCSM: designing safe and potent herbicides with graph-based signatures. Briefings in Bioinformatics, 23(2), bbac042.
20.Rasul, H. O., Ghafour, D. D., Aziz, B. K., Hassan, B. A., Rashid, T. A., & Kivrak, A. (2025). Decoding drug discovery: exploring A-to-Z in silico methods for beginners. Applied Biochemistry and Biotechnology, 197(3), 1453-1503.
21.Seal, S. (2024). Using cell painting and chemical data for small-molecule bioactivity and toxicity prediction. University of Cambridge (United Kingdom).
22.Thu, V. T. A., Nhu, N. Q., Anh, N. T. V., Lim, S. A., Seong, H. J., Md Rasheduzzaman, J., ... & Cho, Y. S. (2025). Deciphering linezolid-induced hematologic toxicity: Targeting TOP2A and TOP2B via its primary metabolite PNU142586. Science advances, 11(22), eadt5833.
23.Tripathi, R. (2025). Decoding Life: Tools and Techniques in Chemical Biology. Chyren Publication.
24.Wang, H., Chen, J., Liu, W., Wang, D., Song, Y., Hong, H., ... & Zimmerman, J. B. (2026). Using machine learning for green substitution of industrial chemicals: Integrating functionality, hazard, and life cycle impact. Chemical Reviews, 126(2), 841-894.
25.Zhang, J., Li, H., Zhang, Y., Huang, J., Ren, L., Zhang, C., ... & Zhang, Y. (2025). Computational toxicology in drug discovery: applications of artificial intelligence in ADMET and toxicity prediction. Briefings in Bioinformatics, 26(5), bbaf533.
