Lichen derived compounds have been reported to have various therapeutic potentials. The present work explores anti-cancer potential of such compounds against human Non-Small Cell Lung Cancer cell line NCI-H322M by using computational methods. Initially a predictive model was developed based on machine learning approach using several classifiers like Random Forest, J48, Bagging, PART and Random Tree in WEKA software. Random tree classifier showed its potency in terms of sensitivity (1), specificity (1), time (.02 sec) and other parameters. This model screened nineteen compounds with active potential out of seventy lichen compounds. Docking simulations were further performed to evaluate the binding potential of those molecules with the tyrosine kinase domain of epidermal growth factor receptor using AutoDock Vina. Four compounds namely Asperphenamate (-9.9 kcal/mol), Brefeldin A(-9.2 kcal/mol), Simvastatin (-10.2 kcal/mol) and Gliotoxin (-8.4 kcal/mol) showed excellent binding potential in the Erlotinib (-8.2 kcal/mol) binding site where as five compounds namely Aculeatins A (-5.6 kcal/mol), Brefeldin A (-6.6 kcal/mol), Compactin (-6.3 kcal/mol), Wortmannin (-9.7 kcal/mol),Phaeosphaerin B (-7.7 kcal/mol )showed their potential in Gefitinib (-7.9 kcal/mol) binding site. Additionally pharmacophores evaluation was done on the screened molecules to compare them with common pharmacophores of Erlotinib and Gafitinib. Results reflect presence of considerable number of heterocycle fragments in the screened compounds with accepted range of other pharmacological properties.
