By using multiple linear regression (MLR) and partial least squares (PLS) regression (models 6–9 in Table 3) (Fu et al., 2012, Chen et al., 2001, Niu et al., 2006, Li et al., 2008 and Fang et al., 2009), both ELUMO and EHOMO were found to play important roles in the QSPR models to predict the photolytic dehalogenation reactivity for PBDEs and PCBs. Ohura et al. (2008) also reported that Eperezolid ELUMO+1, ELUMO and ELUMO − EHOMO are the significant factors affecting photolysis in the QSPR models for chlorinated polycyclic aromatic hydrocarbons.
In addition, artificial neural networks (ANN) are model-free mapping devices antigenic determinant are capable of capturing complex nonlinear relationships in the underlying data that are often missed by conventional QSAR approaches such as MLR and PLS (Karelson et al., 1996 and Du et al., 2008). As a non-linear method, ANN is rapidly becoming the method of choice for structure–activity and structure–property correlations.