Related Experiment Videos
Multi-Objective Genetic Algorithm (MOGA) As a Feature Selecting Strategy in the Development of Ionic Liquids'
Maciej Barycki1, Anita Sosnowska1, Karolina Jagiello1
1Faculty of Chemistry, Department of Environmental Chemistry and Radiochemistry, Laboratory of Environmental Chemometrics , University of Gdansk , ul. Wita Stwosza 63 , 80-308 Gdansk , Poland.
Journal of Chemical Information and Modeling
|December 4, 2018
Summary
Quantitative toxicity-toxicity relationship (QTTR) models predict chemical toxicity across organisms. This study developed a QTTR model for ionic liquids using QSAR data, achieving high accuracy despite limited experimental results.
Area of Science:
- Toxicology
- Computational Chemistry
- Environmental Science
Background:
- Quantitative toxicity-toxicity relationship (QTTR) models offer potential for interpreting toxicological data across different tests and organisms.
- Developing QTTR models is often hindered by limited experimental data.
- Ionic liquids are a class of chemicals with diverse applications and varying toxicological profiles.
Purpose of the Study:
- To develop a QTTR model for predicting the toxicity of ionic liquids.
- To address data scarcity challenges in QTTR model development.
- To evaluate the utility of quantitative structure-activity relationship (QSAR) models in generating data for QTTR development.
Main Methods:
- Utilized quantitative structure-activity relationship (QSAR) models to generate toxicity data for ionic liquids against human cancer cell lines (HeLa and MCF-7).
- Developed a QTTR model using the generated QSAR data.
- Employed a multi-objective genetic algorithm (MOGA) for feature selection in the QSAR and QTTR modeling processes.
Main Results:
- Successfully developed a QTTR model for ionic liquids with a high coefficient of determination (R² = 0.8).
- Demonstrated the feasibility of using QSAR-derived data for QTTR model construction.
- Highlighted the effectiveness of MOGA in selecting optimal features for complex modeling tasks.
Conclusions:
- QTTR models can be effectively developed even with limited experimental data by leveraging QSAR predictions.
- QSAR modeling provides a viable strategy to augment data for building predictive toxicological relationships.
- The developed QTTR model shows promise for predicting ionic liquid toxicity, facilitating risk assessment.