Title: Predicting Toxicity of Ionic Liquid Compounds Using Random Forest Approach

Abstract: Ionic liquids have applications across various scientific fields, in part, due to their interesting physical and chemical properties. Comprehensive assessments of their toxicological profiles are necessary to allow their safe and suitable applications. Unlike prior ionic liquid toxicity predictions which rely on small descriptor sets, we integrate joint 2D and 3D descriptors with random forest which explains the most important descriptors to address these limitations. This study aims to predict the toxicity of ionic liquids using 2D and 3D molecular descriptors by utilizing machine learning. In particular, we propose the random forest regression model to uncover molecular descriptors and toxicity patterns. Additionally, GridSearchCV is used to tune the hyperparameters to ensure optimal model performance. Several metrics were calculated to evaluate the model’s accuracy. The model achieved R²=0.879 which indicates strong predictive performance. Our study demonstrates the benefits of 2D and 3D descriptors for predicting the toxicity of ionic liquids, showing strong correlations between experimental and predicted toxicities. Our analysis of features using 2D and 3D descriptors highlighted those descriptors that are strongly associated with toxicity predictions. Feature importance highlights that physicochemical factors effect toxicity which provides interpretation for ionic liquid design. This study demonstrates the effectiveness of predicting the toxicity of ionic liquids by integrating molecular descriptors and machine learning, thereby facilitating the safer production and application of ionic liquids.

Keywords: Ionic Liquids, Random Forest, Explainable ML, Toxicity Prediction, GridSearchCV, hyperparameter tuning, Molecular Descriptors, 2D and 3D descriptors

• Earle, M. J.; Esperança, J. M.; Gilea, M. A.; Canongia Lopes, J. N.; Rebelo, L. P.; Magee, J. W.; Seddon, K. R.; Widegren, J. A. The distillation and volatility of ionic liquids. Nature 2006, 439, 831–834.
• Wang, Z.; Song, Z.; Zhou, T. Machine learning for ionic liquid toxicity prediction. Processes 2020, 9, 65.
• Mousavi, S. P.; Nakhaei-Kohani, R.; Atashrouz, S.; Hadavimoghaddam, F.; Abedi, A.; Hemmati-Sarapardeh, A.; Mohaddespour, A. Modeling of H2S solubility in ionic liquids: comparison of white-box machine learning, deep learning and ensemble learning approaches. Scientific Reports 2023, 13, 7946.
• Hemmerich, J.; Ecker, G. F. In silico toxicology: From structure–activity relationships towards deep learning and adverse outcome pathways. Wiley Interdisciplinary Reviews: Computational Molecular Science 2020, 10, e1475.
• Goh, G. B.; Siegel, C.; Vishnu, A.; Hodas, N. O.; Baker, N. Chemception: a deep neural network with minimal chemistry knowledge matches the performance of expert-developed QSAR/QSPR models. arXiv preprint arXiv:1706.06689 2017,
• Pretti, C.; Chiappe, C.; Pieraccini, D.; Gregori, M.; Abramo, F.; Monni, G.; Intorre, L. Acute toxicity of ionic liquids to the zebrafish (Danio rerio). Green Chemistry 2006, 8, 238–240.
• Pawłowska, B.; Telesiński, A.; Biczak, R. Phytotoxicity of ionic liquids. Chemosphere 2019, 237, 124436.
• Tabaaza, G. A.; Tackie-Otoo, B. N.; Zaini, D. B.; Otchere, D. A.; Lal, B. Application of machine learning models to predict cytotoxicity of ionic liquids using VolSurf principal properties. Computational Toxicology 2023, 26, 100266.
• Yuan, Q.; Wei, Z.; Guan, X.; Jiang, M.; Wang, S.; Zhang, S.; Li, Z. Toxicity prediction method based on multi-channel convolutional neural network. Molecules 2019, 24, 3383.
• Baskin, I.; Epshtein, A.; Ein-Eli, Y. Benchmarking machine learning methods for modeling physical properties of ionic liquids. Journal of Molecular Liquids 2022, 351, 118616.
• Fan, D.; Xue, K.; Zhang, R.; Zhu, W.; Zhang, H.; Qi, J.; Zhu, Z.; Wang, Y.; Cui, P. Application of interpretable machine learning models to improve the prediction performance of ionic liquids toxicity. Science of The Total Environment 2024, 908, 168168.
• Zhu, P.; Kang, X.; Zhao, Y.; Latif, U.; Zhang, H. Predicting the toxicity of ionic liquids toward acetylcholinesterase enzymes using novel QSAR models. International journal of molecular sciences 2019, 20, 2186.
• Zhou, Z. H. Ensemble methods: foundations and algorithms; CRC Press, 2012.
• Weininger, D. SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules. Journal of chemical information and computer sciences 1988, 28, 31–36.
• Chen, G.; Song, Z.; Qi, Z.; Sundmacher, K. Generalizing property prediction of ionic Liquids from limited labeled data: a one-step framework empowered by transfer learning. Digital Discovery 2023, 2, 591-601.
• Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... & Duchesnay, É. (2011). Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12, 2825-2830.
• Mining, W. I. Data mining: Concepts and techniques. Morgan Kaufmann 2006, 10, 4.
• Guyon, I.; Weston, J.; Barnhill, S.; Vapnik, V. Gene selection for cancer classification using support vector machines. Machine learning 2002, 46, 389–422.
• Schonlau, M.; Zou, R. Y. The random forest algorithm for statistical learning. The Stata Journal 2020, 20, 3–29.
• Bishop, C. M. Pattern recognition and machine learning. Springer Google Scholar 2006, 2, 1122–1128.
• Bergstra, J.; Bengio, Y. Random search for hyperparameter optimization. Journal of machine learning research 2012, 13.
• Willmott, C. J.; Matsuura, K. Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate research 2005, 30, 79–82.
• Chai, T.; Draxler, R. R. Root mean square error (RMSE) or mean absolute error (MAE)?–Arguments against avoiding RMSE in the literature. Geoscientific Model Development, 2014, 7, 1247–1250.
• Cohen, I.; Huang, Y.; Chen, J.; Benesty, J.; Benesty, J.; Chen, J.; Huang, Y.; Cohen, I. Pearson correlation coefficient. Noise reduction in speech processing 2009, 1–4.
• Kvålseth, T. O. (1985). Cautionary Note about R 2 . The American Statistician, 39(4), 279–285.
• Breiman, L. Random forests. Machine learning 2001, 45, 5–32.
• Heuillet, A.; Couthouis, F.; Díaz-Rodríguez, N. Explainability in deep reinforcement learning. Knowledge-Based Systems 2021, 214, 106685.
• Zhao, Yongsheng, et al. "Toxicity of ionic liquids: database and prediction via quantitative structure–activity relationship method." Journal of hazardous materials 278 (2014): 320-329.
• Ahmadi, Shahin, Shahram Lotfi, and Parvin Kumar. "Quantitative structure–toxicity relationship models for predication of toxicity of ionic liquids toward leukemia rat cell line IPC-81 based on index of ideality of correlation." Toxicology Mechanisms and Methods 32.4 (2022): 302-312.
• Peng, Daili, and Francesco Picchioni. "Prediction of toxicity of Ionic Liquids based on GC-COSMO method." Journal of hazardous materials 398 (2020): 122964.
• Chipofya, Mapopa, Hilal Tayara, and Kil To Chong. "Deep probabilistic learning model for prediction of ionic liquids toxicity." International Journal of Molecular Sciences 23.9 (2022): 5258.
• Landrum G. What are the VSA descriptors? RDKit blog. 2023 Apr 17. Available from: https://greglandrum.github.io/rdkit-blog/posts/2023-04-17-what-are-the-vsa-descriptors.html
• Alexander DL, Tropsha A, Winkler DA. Beware of R(2): Simple, Unambiguous Assessment of the Prediction Accuracy of QSAR and QSPR Models. J Chem Inf Model. 2015 Jul 27;55(7):1316-22.
• Sadaghiyanfam, Safa, Hiqmet Kamberaj, and Yalcin Isler. "Enhanced prediction of ionic liquid toxicity using a meta-ensemble learning framework with data augmentation." Artificial Intelligence Chemistry 3.1 (2025): 100087.