Related Experiment Video
Updated: Aug 28, 2025
Advanced 3D Liver Models for In vitro Genotoxicity Testing Following Long-Term Nanomaterial Exposure
Published on: June 5, 2020
Machine learning methods for multi-walled carbon nanotubes (MWCNT) genotoxicity prediction.
Marianna Kotzabasaki1, Iason Sotiropoulos1, Costas Charitidis1
1School of Chemical Engineering, National Technical University of Athens 9 Heroon Polytechneiou Street, Zografou Campus 15780 Athens Greece mariannako@chemeng.ntua.gr hsarimv@central.ntua.gr +30 2107723138 +30 2107723236 +302107723237.
This study developed a predictive nanoinformatics model to assess the genotoxicity of multi-walled carbon nanotubes (MWCNTs). The model accurately predicts MWCNT genotoxicity using key physicochemical properties, offering a faster alternative to experimental testing.
Area of Science:
- Nanomaterials Science
- Computational Toxicology
- Nanoinformatics
Background:
- Multi-walled carbon nanotubes (MWCNTs) possess unique properties driving industrial and biomedical use.
- MWCNT physicochemical characteristics influence their toxicological profiles, necessitating hazard assessment.
- In silico modeling offers a cost-effective alternative to experimental toxicity testing for MWCNTs.
Purpose of the Study:
- To develop and validate a predictive nanoinformatics model for MWCNT genotoxicity.
- To utilize statistical and machine learning approaches for accurate hazard characterization.
- To identify key physicochemical properties influencing MWCNT genotoxicity.
Main Methods:
- Computational workflows combining unsupervised (PCA) and supervised (SVM, RF, LR, NB) learning.
- Bayesian optimization and Recursive Feature Elimination (RFE) for variable selection.
- Development of a Random Forest (RF) model.
Main Results:
- An RF model achieved 80% accuracy in predicting MWCNT genotoxicity on external validation.
- The model identified "Length", "Zeta average", and "Purity" as the most informative features.
- High classification probabilities were achieved, indicating model reliability.
Conclusions:
- A validated nanoinformatics model can accurately predict MWCNT genotoxicity.
- Physicochemical properties like length, zeta average, and purity are crucial for genotoxicity assessment.
- This approach provides a rapid and cost-effective method for MWCNT hazard characterization.
