A QSAR study of integrase strand transfer inhibitors based on a large set of pyrimidine, pyrimidone, and pyridopyrazine carboxamide derivatives

In the present study, 199 compounds derived from pyrimidine, pyrimidone and pyridopyrazine carboxamides with inhibitory activity against HIV-1 integrase were modeled. Subsequently, a multi-variate QSAR study was conducted with 54 molecules employed by Ordered Predictors Selection (OPS) and Partial Least Squares (PLS) for the selection of variables and model construction, respectively. Topological, electrotopological, geometric, and molecular descriptors were used. The selected real model was robust and free from chance correlation; in addition, it demonstrated favorable internal and external statistical quality. Once statistically validated, the training model was used to predict the activity of a second data set (n = 145). The root mean square deviation (RMSD) between observed and predicted values was 0.698. Although it is a value outside of the standards, only 15 (10.34%) of the samples exhibited higher residual values than 1 log unit, a result considered acceptable. Results of Williams and Euclidean applicability domains relative to the prediction showed that the predictions did not occur by extrapolation and that the model is representative of the chemical space of test compounds. (C) 2017 Elsevier B.V. All rights reserved.