Recognizing novel drugs against Keap1 in Alzheimer’s disease using machine learning grounded computational studies

Abstract

Alzheimer’s disease (AD) is the most common neurodegenerative disorder in the world, affecting an estimated 50 million individuals. The nerve cells become impaired and die due to the formation of amyloid-beta (Aβ) plaques and neurofibrillary tangles (NFTs). Dementia is one of the most common symptoms seen in people with AD. Genes, lifestyle, mitochondrial dysfunction, oxidative stress, obesity, infections, and head injuries are some of the factors that can contribute to the development and progression of AD. There are just a few FDA-approved treatments without side effects in the market, and their efficacy is restricted due to their narrow target in the etiology of AD. Therefore, our aim is to identify a safe and potent treatment for Alzheimer’s disease. We chose the ursolic acid (UA) and its similar compounds as a compounds’ library. And the ChEMBL database was adopted to obtain the active and inactive chemicals against Keap1. The best Quantitative structure-activity relationship (QSAR) model was created by evaluating standard machine learning techniques, and the best model has the lowest RMSE and greatest R2 (Random Forest Regressor). We chose pIC50 of 6.5 as threshold, where the top five potent medicines (DB06841, DB04310, DB11784, DB12730, and DB12677) with the highest predicted pIC50 (7.091184, 6.900866, 6.800155, 6.768965, and 6.756439) based on QSAR analysis. Furthermore, the top five medicines utilize as ligand molecules were docked in Keap1’s binding region. The structural stability of the nominated medications was then evaluated using molecular dynamics simulations, RMSD, RMSF, Rg, and hydrogen bonding. All models are stable at 20 ns during simulation, with no major fluctuations observed. Finally, the top five medications are shown as prospective inhibitors of Keap1 and are the most promising to battle AD.