MACHINE LEARNING-ASSISTED DRUG REPURPOSING FOR IDENTIFICATION OF POTENTIAL HDAC6 INHIBITORS

Abstract

Together, the DNA and epigenome tightly regulate neuronal function and differentiation. The abnormal functioning of the genome and epigenome that results from the epigenetic alterations that occur in the face of environmental input leads to neurodegeneration. Histone deacetylases or HDAC constitute a class of proteins or cofactors that need Zn2+ and contribute to the transcription and operation of cells. The overexpression of these proteins, which is prevalent in the development of diverse anomalies in the brain tissues, leads to the dysregulation of several target proteins involved in cell formation and growth associated with Alzheimer's disease, which impairs memory and learning ability. Although several strategies have been used to regulate the greater expression of HDACs using diverse chemical inhibitors, very limited success has been achieved. In the given study we have used machine learning approach to extract drug inhibitor data and target inhibitors. Algorithms such as Random Forest and Support Vector Machine have been used to preprocess data and add required additional parameters like rotatable bonds, canonical smiles, molecular weight, number of atoms, etc. models were trained and evaluation of the models were performed the prediction of data. Eventually molecular docking was done and a list of top 10 novel compounds were retrieved based on their binding affinities with HDAC6. The best binding drug was Bicalutamide, which was an anti-cancerous drug and can be used to treat AD by inhibiting HDAC6.