Using comparative molecular field analysis (CoMFA), a 3D-QSAR model of anticarcinogenicity (inhibiting the induction of ODC) for a series of retinoids was established. The CoMFA model was validated and built by cross-validation (leave-one-out) and non-validation (randomizing) techniques. The significant PLS cross-validated value R2CV (0.905) indicated that the model could be used as a predictive tool for further design of new molecules with high activity. The activities of three compounds excluded from the correlation analysis were computed using this model, small residues obtained. Based on the conformers with the lowest energy, no statistical significance existed (R2CV = 0.405). A performance of minor rotation around a single bond for several compounds provided the conformers with the variation in energy limit less than 2 kcal / mol, by which a correlation analysis was run with a satisfactory relationship between the perturbed conformers and the activities, suggesting that the lowest energy conformers for some compounds be unlikely the pharmacophoric conformers. Superimposition of ligand fields was carried out to mimic the environment of ligands interacting with their receptor, and to visualize the steric and electrostatic behaviors of groups and / or atoms between ligands and receptor. The molecular field model as a template is able to predict activities, and to some extent, to map the topological and physico-chemical characteristics of receptor.
