MIA‐QSAR as an Alternative Approach for Modeling Some Antifungals
Matheus P. Freitas, Roberto Rittner · QSAR & Combinatorial Science · 2007
Abstract A congeneric series of 18 antifungal compounds has been modeled using the Multivariate Image Analysis applied to Quantitative Structure–Activity Relationship (MIA‐QSAR) method, in which descriptors are pixels with non‐zero variance among the 2D molecular structures. Calibration was carried out using PLS and an R2 of 0.956 was obtained for five latent variables. The predictive ability of the model was tested through leave‐one‐out cross‐validation, giving a Q$\rm{ _{CV}^2 }$ of 0.621 after removing two outliers. The model showed to be robust and the results compared reasonably well with a previous biparametric approach, having the potential of be used as an alternative and powerful QSAR approach to model novel actives.