Drug discovery with graph neural networks

Drug discovery is a challenging and complex field characterized by high costs, low success rates, and lengthy timelines. Global health crises, such as the COVID-19 pandemic, have underscored the urgent need for faster, more effective, and innovative solutions in this domain. This thesis aims to accelerate drug discovery processes and reduce associated costs by utilizing Graph Neural Networks (GNNs). GNNs are powerful algorithms capable of analyzing graph structures, represented by nodes and edges, allowing for the modeling of molecular structures and interactions. These algorithms can be effectively applied in various drug discovery applications, including drug-protein binding prediction, drug similarity analysis, scaffold extraction, and the prediction of drug-induced side effects. A particularly promising strategy in this context is drug repurposing, which involves adapting existing approved drugs to new therapeutic applications or using them to treat different diseases. This strategy holds significant potential for providing rapid and effective solutions during emergencies, such as pandemics. GNNs play a crucial role in this repurposing process by quickly and accurately determining the suitability of existing drugs for new targets. The advancement of GNN technology has facilitated pioneering research, especially in Turkey. This thesis exemplifies such pioneering efforts by analyzing molecular structures obtained from resources like DrugBank and PubChem using GNN models to predict drug-protein interactions. The methods employed include atom pair similarity analysis, Tanimoto similarity, and molecular fingerprint techniques. Furthermore, the study focuses on kinase inhibitors, compounds that interact with target proteins to inhibit their activities, playing a critical role in cancer therapy. The findings demonstrate the efficacy of GNNs in drug discovery and repurposing during emergencies like COVID-19 and Ebola, highlighting their potential to expedite the drug discovery process and reduce costs. The importance of further research on the broader application of GNNs in drug discovery and development and the need to keep abreast of technological advancements in this rapidly evolving field are emphasized.