Machine Learning and Molecular Dynamics Based Insights into Mode of Actions of Insulin Degrading Enzyme Modulators

Background: Alzheimer's disease (AD) is one of the most common lethal neurodegenerative disorders having impact on the lives of millions of people worldwide. The disease lacks effective treatment options and the unavailability of the drugs to cure the disease necessitates the development of effectual anti-Alzheimer drugs. Several mechanisms have been reported underlying the association of the two disorders, diabetes and dementia, one among which is the insulin-degrading enzyme (IDE) which is known to degrade insulin as well beta-amyloid peptides. Methods: The present study is aimed to generate accurate classification models using machine learning techniques, which could identify IDE modulators from a bioassay dataset consisting of IDE inhibitors as well as non-inhibitors. The identified compounds were subjected to docking and Molecular dynamics (MD) studies for an in-depth analysis of the binding modes along with the complex stability. This study proposes that the identified potential active compounds, STK026154 (PubChem ID: CID2927418) with Glide score of -7.70 kcal/mol and BAS05901102 (PubChem ID: CID3152845) with Glide score of -7.06 kcal/mol, could serve as promising leads for the development of novel drugs against AD. Conclusion: The present study shows that such in silico approaches can be effectively used to discover and select active compounds from unseen data for accelerated drug development process. The machine learning models generated in the present study were used to screen Traditional Chinese Medicine (TCM) database to identify the phytocompounds already been reported to have therapeutic effects against AD.