A Novel Methodology for Design of Qsar Model: Mutagenicity of Nitroaromatic Compounds

Nitroaromatic compounds (NACs) are omnipresent and pervasive organic pollutants. Here, we aim to design transparent, interpretable, reproducible, and publicly available methodologies for deriving quantitative structure-activity relationship models and testing them by modeling the mutagenicity of NACs against the Salmonella typhimurium TA100 strain. Selection of molecular descriptors are performed using a genetic, Boruta, Featurewiz, and ForwardSelector algorithm. Modelling utilizes the multiple linear regression method. The models are ranked by the Multi-Criteria Decision Making procedure. Our findings reveal that Featurewiz showing a slight advantage over Boruta and ForwardSelector and significantly outperforms genetic algorithm. Constructed models can serve as valuable tools for the quick and reliable prediction of NACs mutagenicity.