A Statistical Comparison between Zagreb indices for correlation with toxicity predictions of natural products

Abstract

Graph theory had wide applications in developing in silico tools and it is widely used to calculate topological indices to establish structural activity relations of chemicals/compounds. However, usage of Zagreb indices with respect to natural compounds activity/toxicity prediction needs more attention. Many available online tools are using atom bond connectivity index (ABC Index), first and second Zagreb indices. The usage of the Hyper Zagreb index is very rare and using natural compounds is neglected. In this context, three types of Zagreb indices (first Zagreb index, second Zagreb index and hyper Zagreb index) were calculated to the selected chemical compounds of natural products and the relation between these indices and cytotoxicity of natural compounds were established. We have selected IC50 Values of the selected natural compounds in Hela cell lines as an index for cytotoxicity from the literature. The correlation of Zagreb indices and activity was performed using the R program, and we reached the conclusion that all indices correlate with the cytotoxicity of the studied compounds. This study acts as evidence to prove that, hyper Zagreb index correlates more with the cytotoxicity/activity of the studied natural compounds. Further studies using other Machine Learning tools to verify these findings will establish the importance of the hyper Zagreb index as one method to predict the toxicity of natural compounds.