INTEGRATIVE COMPUTATIONAL APPROACHES FOR RECEPTOR-BASED DRUG DISCOVERY AND BIOMARKER IDENTIFICATION IN NDDs

Abstract

Neurodegenerative disorders (NDDs) represent a significant global health burden, with overlapping clinical and molecular features that often complicate early diagnosis and treatment. This thesis presents an integrative computational approach to address key challenges in the therapeutic and diagnostic landscape of NDDs by bridging receptor-based drug discovery and gene expression-driven biomarker identification. The first objective focuses on targeting the GABRA2 receptor, a subunit of the GABA A complex implicated in both anxiety and neurodegeneration. Using homology modelling, ligand-based virtual screening, molecular docking, and ADME profiling, the study identifies novel FDA-approved compounds with improved binding affinity and pharmacokinetic properties over the benchmark drug Diazepam. Zolmitriptan emerged as a promising candidate with high BBB permeability, favourable bioavailability, and minimized toxicity. The second objective employs machine learning models on high-throughput transcriptomic data (GSE140830) to classify dementia subtypes and identify key biomarkers. Random Forest, Support Vector Classifier, and other models achieved robust classification performance, while feature importance and pathway enrichment analyses revealed subtype-specific gene signatures linked to neuroinflammatory and synaptic pathways. By unifying molecular pharmacology with ML-driven omics analytics, this study provides a dual-framework for stratified therapeutic targeting and early diagnosis in NDDs, offering translational value for precision medicine.