Design and screening of HepG2 cancer cell line inhibitors from Triterpenoid derivatives of Paramignya Trimera

Currently, Artificial intelligence (AI) is a ubiquitous technology that provides effective support across all fields. The pharmaceutical industry in general and drug production and development, in particular, are enjoying a very good application for the opportunity when in silico models have emerged as powerful platforms for designing new drugs. The aim of this project is to develop new anti-cancer agents by designing novel Triterpenoid derivatives from Paramignya Trimera and predicting their efficacy against the Bcl-2 target receptor. The project used three main in silico models: QSARMLR, QSARPCR and QSARANN. The models can be used to estimate IC50 values for novel derivatives and Escin extracted from Paramignya Trimera. Finally, the new good-value derivatives were docked to the Bcl-2 receptor to assess responsiveness. As a result, newly designed 196 compounds from the structural framework of Triterpenoid compounds were designed by combined with potential substituents. From there, screening by the rule of Veber identified 138 substances that met the requirement of having the ability to make drugs. Successfully, built QSARMLR, QSARPCR, QSARANN models with results of statistical values: R2 = 0.849, R2adj = 0.826, Q2LOO = 0.789 for the QSARMLR model; QSARPCR model with R2 = 0.860, R2adj = 0.831, Q2LOO = 0.805, and the QSARANN model with the best results: R2train = 0.941, R2test = 0.915, R2cv = 0.912. The use of models can help predict the effectiveness of newly engineered compounds. In this study, 20 compounds were found to be more efficient than Escin. Molecular docking on the Bcl-2 receptor found T.new7 gave the most potential results with the binding energy E_binding = -7.933 (kcal.mol-1), RMSD = 1.915 (Å). The research has achieved its goal by finding T.new7, a newly designed compound with better anti-cancer ability than natural Escin.