Prediction model based on decision tree analysis for laccase mediators

A Structure Activity Relationship (SAR) study for laccase mediator systems was performed in order to correctly classify different natural phenolic mediators. Decision Tree (DT) classification models with a set of five quantum-chemical calculated molecular descriptors were used. These descriptors included redox potential, ionization energy, pKa, enthalpy of formation of radical and O-H bond dissociation energy (DO-H). The rationale for selecting these descriptors is derived from the laccase-mediator mechanism. To validate the DT predictions, the kinetic constants of different compounds as laccase substrates, their ability for pesticide transformation as laccase-mediators, and radical stability were experimentally determined using C. gallica laccase and the pesticide dichlorophen. The prediction capability of the DT model based on three proposed descriptors showed a complete agreement with the obtained experimental results.