Deep Learning Unravels Potential Antibiotic-Resistance Drugs against Klebsiella Pneumoniae

Motivation: Deep learning (DL) techniques are utilized to accelerate drug discovery and minimize risks during clinic trials for patients infected with Klebsiella Pneumoniae (KP) infections, enabling faster and more effective recovery. Results: Key data such as targets, enzymes, SMILES, and pathways were extracted from Drug-Bank for 3,475 selected biotech and small molecule drugs. A convolutional neural network (CNN) incorporating drug-drug interaction (DDI) data was used for training and prediction, achieving over 70% an accuracy. Additionally, evolutionary scale modeling (ESM) identified five promising drugs with molecular similarities exceeding 85% to KP strains collected from hospital datasets.