Topological QSAR Modelling of Carboxamides Repellent Activity to Aedes Aegypti

Abstract

Aedes aegypti vector control is of paramount importance owing to the damages induced by the various severe diseases that these insects may transmit, and the increasing risk of important outbreaks of these pathologies. Search for new chemicals efficient against Aedes aegypti, and devoid of side-effects, which have been associated to the currently most used active substance i. e. N,N-diethyl-mtoluamide (DEET), is therefore an important issue. In this context, we developed various Quantitative Structure Activity Relationship (QSAR) models to predict the repellent activity against Aedes aegypti of 43 carboxamides, by using Multiple Linear Regression (MLR) and various machine learning approaches. The easy computation of the four topological descriptors selected in this study, compared to the CODESSA descriptors used in the literature, and the predictive ability of the here proposed MLR and machine learning models developed using the software QSARINS and R, make the here proposed QSARs attractive. As demonstrated in this study, these models can be applied at the screening level, to guide the design of new alternatives to DEET.

Machine Learning Models and New Computational Tool for the Discovery of Insect Repellents that Interfere with Olfaction

“…Unlike the interactions with the olfactory system, repellency interactions have already been modeled [26,27,37,[74][75][76][77][78][28][29][30][31][32][33][34][35], but with in-house systems. Thus, we reduced to select the most commonly used chemotypes IRs, like carboxamides and compounds with botanical origin.…”

“…This model used three MDs and was trained with the Vote metaclassi er. [33,36]. However, QuBiLS-MIDAS software allowed improving the results of QSAR modeling as shown in the statistical performance of the four different datasets used here.…”

“…Quantitative Structure-Activity Relationships (QSAR) modeling approach attempted successfully for drug discovery and has been widely used for search new repellents [26,27,36,37,[28][29][30][31][32][33][34][35]. However, several studies are based on congeneric data, that is, data from families of related compounds.…”

“…but differing in the software used to calculate the MDs. Models with exceptional stability and predictive ability were shaped in both cases[32,33]. Janairo et al used 20 DFT calculated MDs to predict the repellency of 33 plant-derived compounds[34].…”