Predicting the preferred conformations of luteolin-4'-O-β-D-glucoside in gas phase: a comparison of two computational approaches.

A tree-step computational approach has been applied to determine the lowest-energy conformers of luteolin-4'-O-β-D-glucoside (L4'G). Fifty-seven starting structures of the L4'G have been built, and then by performing with density functional theory (DFT) optimizations and second-order Møller-Plesset (MP2) calculations, the preferred conformations of L4'G are predicted. In order to test the accuracy of the computational approach, a hybrid Monte-Carlo multiple minimum (MCMM)/quantum mechanical (QM) approach is applied to determine the favorable conformers of L4'G. The alternative classification is employed to put similar conformations into the same catalogue according to the dihedral angles among the luteolin rings, glycosidic dihedral angles, and the orientations of hydroxyl and hydroxymethyl groups. The low-energy conformations are located after the optimizations at the HF/6-31G(d) and B3LYP/6-311+G(d) levels. Compared with the hybrid MCMM/QM approach, the tree-step computational approach not only remains accurate but also saves a lot of computing resources.