Structure–carcinogenic activity relationship studies of polycyclic aromatic hydrocarbons (PAHs) with pattern-recognition methods

Abstract

Recently a new methodology to identify the carcinogenic activity of polycyclic aromatic hydrocarbons (PAHs) was proposed. This methodology named electronic indices methodology (EIM) is based on the use of local density of states (LDOS) calculations. In this work we perform a comparative study of this methodology with principal component analysis (PCA) and artificial neural networks (ANN). All the physicochemical descriptors were calculated from the molecular eigenstates/spectra obtained through the well-known semi-empirical method parametric method 3 (PM3). PCA and ANN results show that EIM descriptors are relevant to identify the carcinogenic activity of methylated and non-methylated PAHs. Also, we show that the combined use of these distinct methodologies can be an efficient and powerful tool in the structure–activity studies of PAHs or other organic compounds. We have studied 81 methylated and non-methylated PAHs, and our study shows that with the use of these methods it is possible to predict correctly the PAHs' carcinogenic activity with high accuracy (∼80%).