Structure-activity relationship studies of carcinogenic activity of polycyclic aromatic hydrocarbons using calculated molecular descriptors with principal component analysis and neural network methods.

Abstract

Recently a new methodology based on local density of state (LDOS) calculations using topological and semiempirical methods was proposed to identify the carcinogenic activity of polycyclic aromatic hydrocarbons (PAHs). In this work we perform a comparative study of this methodology with principal component analysis (PCA) and neural networks (NN). The PCA and NN results show that LDOS quantum chemical descriptors are relevant descriptors to identify the carcinogenic activity of methylated and non-methylated PAHs. Also, we show that the combination of these distinct methodologies can be an efficient and powerful tool in the structure-activity studies of PAHs 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 correctly predict the carcinogenic activity of PAHs with accuracy higher than 80%.