Development of Predictive Linear and Non-linear QSTR Models for Aliivibrio Fischeri Toxicity of Deep Eutectic Solvents

Development of Predictive Linear and Non-linear QSTR Models for Aliivibrio Fischeri Toxicity of Deep Eutectic Solvents

Amit Kumar Halder (University of Porto, Porto, Portugal) and M. Natália D. S. Cordeiro (LAQV-Requimte, University of Porto, Porto, Portugal)
DOI: 10.4018/IJQSPR.2019100104
OnDemand PDF Download:
No Current Special Offers


Deep eutectic solvents (DESs) have emerged as a very important group of chemicals in recent years. Although generally considered as environment friendly or ‘green,' recent investigations reported toxic behaviors of some DESs towards various biological species. In this work, quantitative structure toxicity relationship analysis was performed on a dataset containing 72 DESs and their components to find the structural determinants responsible for higher DES mediated toxicity. Additionally, efficiencies of various machine learning tools as well as different feature selection algorithms were estimated. To understand the true predictivity of the derived models, three external validation strategies, namely ‘points out,' ‘mixtures out,' and ‘compounds out' were applied along with an ‘all out' technique, a modification of the earlier reported ‘everything out' validation. The models highlight importance of the number of nitrogen atoms, the van der Waals surface area, molar refractivity, dipole moment and molecular mass for shaping the toxicity of DESs and their components towards A. fischeri.
Article Preview


Deep eutectic solvents (DESs) have emerged in recent years as promising environment-friendly alternatives to conventional solvents due to their numerous chemical and industrial applications, especially in the fields of chemistry, materials science, chemical engineering and pharmaceuticals (Cunha & Fernandes, 2018; Francisco, van den Bruinhorst, & Kroon, 2013; Smith, Abbott, & Ryder, 2014). Belonging to the category of low transition temperature mixtures, binary DESs are most commonly prepared by combining a quaternary ammonium salt, which acts as hydrogen bond acceptor, with a suitable hydrogen bond donor (HBD) in a specific molar ratio to produce a eutectic mixture, the melting point of which is less than either of its components (Abbott, Boothby, Capper, Davies, & Rasheed, 2004; Kudlak, Owczarek, & Namiesnik, 2015). DESs share many common characteristics with their predecessor ionic liquids (ILs) such as non-volatility, high conductivity, thermal stability, and solvation properties. Nevertheless, DESs offer many advantages over ILs as most of these DESs are biodegradable, easy to prepare and therefore less expensive (Płotka-Wasylka, Rutkowska, Owczarek, Tobiszewski, & Namieśnik, 2017; Shishov, Bulatov, Locatelli, Carradori, & Andruch, 2017; Smith et al., 2014). It is precisely the environment-friendly or ‘green’ nature of DESs that has drawn enormous attention from the scientific community. In fact, DESs are thought to be less toxic as compared to most conventional organic solvents as well as to their predecessors ILs (Bubalo, Radosevic, Redovnikovic, Halambek, & Srcek, 2014; Das & Roy, 2013; I. Juneidi, Hayyan, & Mohd Ali, 2016). However, some recent investigations have casted doubts on the ‘non-toxic’ or ‘green’ nature of DESs, since it has been observed that some of them may elicit potential toxicity towards various mammalian and microbial cell lines (Hayyan, Hashim, Al-Saadi, et al., 2013; Hayyan, Hashim, Hayyan, et al., 2013; Hayyan, Looi, Hayyan, Wong, & Hashim, 2015; Hayyan et al., 2016; Ibrahim Juneidi, Hayyan, & Hashim, 2015; I. Juneidi et al., 2016). There is thus the need of investigate in detail the toxic adverse effects of DESs. The European Union REACH (Registration, Evaluation and Authorization of Chemicals) legislation envisages the risk assessment of all new and existing chemicals. REACH even encourages the application of Quantitative Structure-Activity Relationships (QSAR) models for a fast and low-cost identification of potential chemical hazards (Worth et al., 2007). For the last six decades, QSAR models have been extensively employed in environmental chemistry, toxicology, pharmaceutical chemistry, and in many other research fields. The major purpose of any QSAR modeling approach is to develop predictive mathematical models to correlate the biological responses or properties of chemicals with their structural information(s) and/or physicochemical properties (Cherkasov et al., 2014; Lewis & Wood, 2014). Even though several chemometric analyses were reported over the past decades to justify the environmental toxicity of ILs, chemometric modeling of DES toxicity is rarely found in the literatures (Das & Roy, 2013; Das et al., 2016; Paterno, Bocci, Goracci, Musumarra, & Scire, 2016). The limited number of available experimental toxicity data of DESs is undoubtedly one of the fundamental barriers for the development of reliable chemometric models of these chemicals. Besides these, the structural natures of DESs, which are more complex than ILs and many other organic compounds, pose serious challenges related not only with the setup of conventional chemometric models but also with the statistical validation and interpretations of models. This work aims at setting up validated Quantitative Structure-Toxicity Relationship (QSTR) models for predicting the experimental toxic responses of a series of DESs and their components against marine bacteria Aliivibrio fischeri. We attempted also to understand possible mechanisms of DESs for their toxic responses towards this microbial organism. However, the scope of the current work may also be extended to address the issues related to descriptor calculation, model development and model validation of small datasets of chemical mixtures.

Complete Article List

Search this Journal:
Open Access Articles
Volume 7: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 6: 4 Issues (2021)
Volume 5: 4 Issues (2020)
Volume 4: 4 Issues (2019)
Volume 3: 2 Issues (2018)
Volume 2: 2 Issues (2017)
Volume 1: 2 Issues (2016)
View Complete Journal Contents Listing