Molecular Docking Using Graph Neural Networks

• Grant number: 22/13156-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2022
• End date: November 30, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: João Paulo Papa
• Grantee: Nicolas Barbosa Gomes
• Host Institution: Faculdade de Ciências (FC). Universidade Estadual Paulista (UNESP). Campus de Bauru. Bauru , SP, Brazil
• Associated research grant: 14/12236-1 - AnImaLS: Annotation of Images in Large Scale: what can machines and specialists learn from interaction?, AP.TEM

Abstract

Medicines act on the human body by modifying the behavior of specific proteins, and the discovery of new compounds to improve the quality and duration of human life is very costly and time-consuming. However, with the help of computational methods and Graph Neural Networks (GNNs), the analysis of the protein-drug interaction can be performed efficiently and quickly, predicting which molecules would have the most significant potential to become a drug. In this aspect, graphs are structures that allow the computer to represent abstract entities of daily life. With the help of these structures, molecules can be represented, considering their biochemical characteristics in three-dimensional space, in a way that the computer can work precisely with the information and interactions between its atoms. If the knowledge of graphs is combined with the comprehension of machine learning, it is possible to infer the interaction and affinity between a protein and a ligand. By understanding these interactions, we can discover the medicinal characteristic of specific molecules since the bonds between proteins and their ligands play an essential role in cellular processes. Thus, the model used in the research aims to train GNNs with a public database to perform molecular docking, that is, find the best fit to anchor the ligand molecule to the protein molecule. Moreover, based on the resulting compound, the classification is made, characterizing whether the existing bond, in this case, has stability or not. Some GNN variants will be considered for this task, and the proposal also comprises an internship period at the University of Wolverhampton, UK.