Title:Multi-Target QSAR Approaches for Modeling Protein Inhibitors. Simultaneous Prediction of Activities Against Biomacromolecules Present in Gram-Negative Bacteria
VOLUME: 15 ISSUE: 18
Author(s):Alejandro Speck-Planche and M.N.D.S. Cordeiro
Affiliation:REQUIMTE/Department of Chemistry and Biochemistry, University of Porto, 4169-007 Porto, Portugal.
Keywords:Amino acid sequence, Inhibitor, Molecular fragment, Moving average approach, Mt-QSAR, Physicochemical
properties.
Abstract:Drug discovery is aimed at finding therapeutic agents for the treatment of
many diverse diseases and infections. However, this is a very slow an expensive process,
and for this reason, in silico approaches are needed to rationalize the search for
new molecular entities with desired biological profiles. Models focused on quantitative
structure-activity relationships (QSAR) have constituted useful complementary tools
in medicinal chemistry, allowing the virtual predictions of dissimilar pharmacological
activities of compounds. In the last 10 years, multi-target (mt) QSAR models have
been reported, representing great advances with respect to those models generated from classical approaches. Thus, mt-
QSAR models can simultaneously predict activities against different biological targets (proteins, microorganisms, cell
lines, etc.) by using large and heterogeneous datasets of chemicals. The present review is devoted to discuss the most
promising mt-QSAR models, particularly those developed for the prediction of protein inhibitors. We also report the first
multi-tasking QSAR (mtk-QSAR) model for simultaneous prediction of inhibitors against biomacromolecules (specifically
proteins) present in Gram-negative bacteria. This model allowed us to consider both different proteins and multiple
experimental conditions under which the inhibitory activities of the chemicals were determined. The mtk-QSAR model
exhibited accuracies higher than 98% in both training and prediction sets, also displaying a very good performance in the
classification of active and inactive cases that depended on the specific elements of the experimental conditions. The
physicochemical interpretations of the molecular descriptors were also analyzed, providing important insights regarding
the molecular patterns associated with the appearance/enhancement of the inhibitory potency.
