Fuzzy Handling of Uncertainties in Modeling the Inhibition of Glycogen Synthase Kinase-3 by Paullones

We consider the problem of modeling the biological activity of inhibitors (belonging to the paullone family) for Glycogen Synthase Kinase-3 (GSK-3). The development of a nonlinear model that establishes the quantitative structure-activity relationship (QSAR) involves an uncertainty regarding the optimal choice of molecular descriptors, structure of the model, number of free model parameters, and so on. This study advocates the use of a fuzzy filter, that mathematically takes into account the underlying uncertainties, for the handling of the uncertainties. The modeling performance is prevented from being adversely affected by the uncertainties with the help of a fuzzy filter. We demonstrate that a robustness of the modeling performance against uncertainties could be achieved using the proposed approach.