Molecular modelling of antiproliferative inhibitors based on SMILES descriptors using Monte-Carlo method, docking, MD simulations and ADME/Tox studies

Abstract

ABSTRACT Cancer is one of the greatest challenges that worry the minds of scientists and threatens human life. Despite the presence of several drugs on the market, their effectiveness remains limited by its resistance. In this research, the Monte Carlo approach was used for QSAR modelling applying the representation of the molecular structure by the SMILES and optimal molecular descriptors. Correlation Ideality (IIC) and Correlation Contradiction Index (CCI)) were introduced as validation parameters to further estimate the predictive ability of the developed models. The statistical quality of the model developed with (IIC) was good compared to those without (IIC). The best QSAR model of the following statistical parameters: (R²train= 0.816, R²valid = 0.825) was selected to generate the activity-increasing and decreasing promoters studied, and these are the basis for the design of seven new candidates as an antiproliferative inhibitory agent. Additionally, the newly designed molecules, the most active compound in the dataset, erlotinib and melphalan as control compounds were docked in the EGFR receptor binding site. The docking results discovered that the proposed candidates had the highest potential and energy affinity. Besides, Lipinski’s Rule of Five, Synthetic Accessibility and ADME/Tox were performed to assess bioavailability and drug-likeness of proposed compounds. In addition, MD simulation accompanied by MM-PBSA analysis discovered that compound A1 and the screened compounds were stable and did not show significant fluctuations throughout the simulation time. Generally, this research showed that the selected model well explains the antiproliferative activity and also that the proposed compounds have high activity, good binding affinity and stable conformation with the reported target protein.