QSAR for non-nucleoside inhibitors of HIV-1 reverse transcriptase.

By means of QSAR algorithms we model the potency pIC(90) [mM] of 154 non-nucleoside reverse transcriptase inhibitors (NNRTI) of the wild-type HIV-1 virus, considered as the second generation analogues of Efavirenz. In addition, 56 inhibitors of the K-103N viral mutant form are also investigated. A pool of 1494 theoretical molecular descriptors provided mainly by the Dragon 5 software is explored by several methods of variable selection: forward stepwise regression, the replacement method, and the genetic algorithm approach. The optimal models found include up to seven parameters: R = 0.7991, R(l-20%-o) = 0.7233 for the case of wild-type, and R = 0.9261, R(l-5%-o) = 0.8802 for the K-103N mutation.