Machine Learning-Driven Insights: Rare Disease Drug Discovery and Cancer Patient Stratification

Abstract

The computational prowess of machine learning (ML), synergized with the establishment and refinement of diverse databases, emerges as a tool to revolutionize drug discovery and cancer treatment. In this thesis, a pipeline incorporating a novel deep-learning based ML method with traditional molecular docking was developed to accelerate drug screening for leishmaniasis. This thesis then concentrated its analytical lens on the categorization of triple negative breast cancer (TNBC) by applying unsupervised learning algorithms on gene expression data. The resultant subtypes, as delineated by each algorithm, underwent a comparative analysis. Furthermore, these clusters were enriched for gene ontology (GO) terms to unveil the distinct characteristics and prognosis implications of potential new clusters.