Establishment of interpretable cytotoxicity prediction models using machine learning analysis of transcriptome features

https://doi.org/10.1016/j.apsb.2025.02.009

This new article publication from Acta Pharmaceutica Sinica B, discusses establishment of interpretable cytotoxicity prediction models using machine learning analysis of transcriptome features.

Cytotoxicity, usually represented by cell viability, is a crucial parameter for evaluating drug safety in vitro. Accurate prediction of cell viability/cytotoxicity could accelerate drug development in the early stage. In this study, by integrating cellular transcriptome and cell viability data using four machine learning algorithms (support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM)) and two ensemble algorithms (voting and stacking), highly accurate prediction models of 50% and 80% cell viability were developed with area under the receiver operating characteristic curve (AUROC) of 0.90 and 0.84, respectively; these models also showed good performance when utilized for diverse cell lines. Concerning the characterization of the employed Feature Genes, the models were interpreted, and the mechanisms of bioactive compounds with a narrow therapeutic index (NTI) can also be analyzed. In summary, the models established in this research exhibit superior capacity to those of previous studies; these models enable accurate high-safety substance screening via cytotoxicity prediction across cell lines. Moreover, for the first time, Cytotoxicity Signature (CTS) genes were identified, which could provide additional clues for further study of mechanisms of action (MOA), especially for NTI compounds.

Keywords: Interpretable model; Drug safety; Cell viability; Weak cytotoxicity; Machine learning; Transcriptome; Cytotoxicity Signature genes; Narrow therapeutic index drugs

Graphical Abstract: available at https://ars.els-cdn.com/content/image/1-s2.0-S2211383525000528-ga1_lrg.jpg

This study provides highly accurate models for 50% and 80% cell viability prediction and the key features which could facilitate the MOA study of substance with a narrow therapeutic index.

# # # # # #

The Journal of the Institute of Materia Medica, the Chinese Academy of Medical Sciences and the Chinese Pharmaceutical Association.

For more information please visit https://www.journals.elsevier.com/acta-pharmaceutica-sinica-b/

Editorial Board: https://www.journals.elsevier.com/acta-pharmaceutica-sinica-b/editorial-board

APSB is available on ScienceDirect (https://www.sciencedirect.com/journal/acta-pharmaceutica-sinica-b).

Submissions to APSB may be made using Editorial Manager® (https://www.editorialmanager.com/apsb/default.aspx).

CiteScore: 22.4

Impact Factor: 14.8 (Top 5 journal in the category of Pharmacology and pharmacy)

JIF without self-citation: 13.9

ISSN 2211-3835

 # # # # #

You Wu, Ke Tang, Chunzheng Wang, Hao Song, Fanfan Zhou, Ying Guo, Establishment of interpretable cytotoxicity prediction models using machine learning analysis of transcriptome features, Acta Pharmaceutica Sinica B, Volume 15, Issue 3, 2025, Pages 1344-1358, ISSN 2211-3835, https://doi.org/10.1016/j.apsb.2025.02.009