Abstract (EN):
Ionic liquids (ILs) possess a unique physicochemical profile providing a wide range of applications. Their almost limitless structural possibilities allow the design of task-specific ILs. However, their "greenness,"specifically their claimed relative nontoxicity has been frequently questioned, hindering their REACH registration processes and, so, their final application. Since the most of ionic liquids has yet to be synthesized, the development of chemoinformatic tools allowing the efficient profiling of the hazardous potential of these compounds becomes critical. In this sense, the combined use of multiple base classifiers (ensembles or multiclassifiers) have proved to overcome the classification performance limitations associated to the use of single classifiers. In the present work we report two ensembles models with good predictive capabilities in a validation set of ionic liquids never used in the learning process. These models were obtained as part of Quantitative Structure Activity Relationship's studies (QSAR) applied to the characterization of neurotoxic profile of ionic liquids based on its inhibition of the Acetyl cholinesterase enzyme (AChE) as neurotoxicity indicator. The results obtained show that one can expect that at least 96% of a set of news ionic liquids can be correctly classified using theses ensembles models. Consequently, these chemoinformatics models provides efficient decision making tools in the design and development of new "green" ionic liquids.
Language:
Spanish
Type (Professor's evaluation):
Scientific
No. of pages:
10