Chemometric Modeling of Algal Toxicity
Melek Türker Saçan, Serli Önlü, Gulcin Tugcu · 2021
The continuous growth of the chemical industry leads to the production and use of an extensive number of chemicals all over the world. Along with their benefits, these chemicals can also have adverse environmental effects, particularly on aquatic ecosystems, such as algal communities. Algae are primary producers of nutrients and oxygen for higher trophic levels, thus, of fundamental importance in ecotoxicity evaluations. Chemometric modeling is one of the alternative methods used for the assessment of adverse effects of chemicals. Quantitative structure–activity relationship (QSAR) is a typical application of chemometric modeling widely used in aquatic toxicity assessment. We reviewed multiple linear regression (MLR)-based QSAR models generated for different algal species in the scientific literature. While some of these linear models are general models, others are developed for some specific class of chemicals. Knowing that the quality of experimental toxicity data is one of the most important criteria in QSAR modeling, we focused on QSAR models generated using data obtained from a standard test protocol. In addition, state-of-the-art internal and external validation metrics, a well-defined applicability domain (AD), and application of the generated models to an external set of chemicals with no experimental data were other criteria considered in the review of MLR-based QSAR models. The importance of a well-predicted toxicity value of a chemical for filling the data gap, screening, and prioritization was emphasized. Moreover, whether further testing is necessary to make a better contribution to hazard assessment was emphasized.