In silico de novo Design of NNRTIs of HIV-1: Functional Group Based Computational Molecular Modeling Approach
Seven novel lead compounds, acting as NNRTIs of HIV-1, are extracted from a database of, in silico de novo designed, 500 compounds. Functional group based computational molecular modelling techniques are used for such design of Acylthiocarbamate derivatives. Effect of structural characteristics on the antiviral activity of these derivatives has also been studied. Statistical regression techniques namely, Non-linear (Back Propagation Neural Network, Support Vector Machine) and linear (Multiple Linear) chemometric regression methods are used in developing the relationships of Kier-Hall Electrotopological State Indices (ERingA, EO8, EN9, EO14, ES16, EN17, EO19, ER, and ER1) with the HIV-1 antiviral activity. The results of assessment of relative potentials of these methods suggest that BPNN (r2 = 0.845, MSE = 0.142, q2 = 0.818) describes the relationship between the descriptors and antiviral activity in a better manner than SVM-ε-radial (r2 = 0.844, MSE = 0.144, q2 = 0.807) and MLR (r2 = 0.836, MSE = 0.150, q2 = 0.805).
De Novo Design, Molecular Modeling, NNRTIs; Back Propagation Neural Networks (BPNN); Support Vector Machine (SVM); Multiple Linear Regression (MLR)
Full Text: PDF (downloaded 100 times)
- There are currently no refbacks.