Multivariate image analysis applied to QSAR: Evaluation to a series of potential anxiolytic agents

A recently implemented QSAR method, whose descriptors derive from bidimensional images, is applied here to model some compounds with anxiolytic activities. A calibration set composed by 47 drug-like molecules (5-HT2C receptor antagonists) was built and a leave-one-out cross-validation, as well as an external validation, was carried out in order to test the predictive ability of the calibration model. The results (R2 = 0.866 and Q2 for the external validation set = 0.681, using 6 latent variables) were comparable to the ones obtained elsewhere, where a CoMFA model was used as 3D approach.