Gramatica, P.; Papa, E. An Update of the BCF QSAR Model Based on Theoretical Molecular Descriptors. QSAR Comb. Sci. 2005, 24, 8, 935–960.

Abstract

An inspection of an already developed QSAR model is necessary to verify the reproducibility of the model itself, and its present applicability to new chemicals, if a version of calculation software for theoretical molecular descriptors is updated. A QSAR model for bioconcentration in fish, published by the authors and based on 5 molecular descriptors of different typology calculated by an old version of DRAGON software, now needs updating and redevelopment owing to changes in some of the descriptor values in the new version of the same package. Here we propose a new OLS model, mechanistically interpretable, based on structural descriptors of the same typology as the previously published QSAR, highlighting the ability of the Genetic Algorithm to select structural information really related to BCF (Bioconcentration Factor). Model predictivity is still guaranteed by statistical internal and external validations; the chemical domain of applicability, verified by the leverage approach, is very similar to that of our previous model.