MoleculeXpert: A Novel Architecture for Expert-Level Molecule Analysis of HIV Inhibition

The global HIV/AIDS pandemic persists as a formidable health challenge, demanding innovative solutions for accelerated drug discovery. In response to this imperative, an advanced platform at the intersection of artificial intelli-gence and healthcare, MoleculeXpert is introduced. Leveraging Graph Neural Networks (GNNs), MoleculeXpert accelerates the identification of potential HIV inhibitors by transforming complex molecular structures into interpretable graph representations. The paper unfolds with a robust framework that combines classification, generative modeling, and explainable AI to streamline drug discovery processes. Utilizing PyTorch Geometric for dataset creation and a Graph Transformer Network for classification, MoleculeXpert classifies molecules as suitable or unsuitable for HIV inhibition. The generative model introduces a pioneering approach, creating novel molecules with potential inhibitory properties. MoleculeXpert's user-friendly interface, developed with Streamlit, ensures accessibility, transparency, and seamless interaction. The incorporation of Explainable AI techniques on graphs enhances the interpretability of results, fostering trust in the generated insights. The dataset, derived from the MoleculeNet HIV dataset via DeepChem, forms the basis for training and testing the model.