Abstract
Quantitative structure-activity relationship analysis of some recently synthesized thiourea derivatives was studied for their cytotoxicity data (CC50) in MT-4 cells. The statistically significant 2D-QSAR model-1 (r2 = 0.964, r2 CV = 0.943 and r2 -pred = 0.961) constructed by genetic function approximation (GFA) methodology was obtained when the number of descriptors was set to 4, indicating that the descriptors of the lowest unoccupied orbital energy (LUMO), radius of gyration (RadOfGyration) and E-state indices (S_ssCH2 and S_aaCH) crucially contributed to biological activity. The validation of the model was done by full cross-validation tests, randomization tests and external test set prediction. Molecular field analysis (MFA) on the corresponding training set investigated the substitutional requirements for the favorable receptor-drug interaction using genetic partial least squares (G/PLS) method, showing that both steric and electrostatic fields play a major role in determining activity. The results derived by combining both methodologies may be useful in further prediction of cytotoxity data (CC50) of corresponding anti-HIV drugs.
Keywords: Genetic function approximation, Molecular field analysis, Quantitative structure-activity relationship, Anti-HIV activity, Thiourea derivatives
2