Skittles: GNN ‐Assisted Pseudo‐Ligands Generation and Its Application for Binding Sites Classification and Affinity Prediction
Year: 2025
Type: article
Abstract: Nowadays, multiple solutions are known for identifying ligand-protein binding sites. Another important task is labeling each point of a binding site with the appropriate atom type, a process known as pseudo-ligand generation. The number of solutions for pseudo-ligand generation is limited, and, to our knowledge, the influence of machine learning techniques has not been studied previously. Here, we describe Skittles, a new graph neural network-assisted pseudo-ligand generation approach, and compare it with known force-field-based methods. We also demonstrate the application of Skittles-based data for solving several important problems in structural biology, including ligand-protein binding site classification and ligand-protein affinity prediction.
DOI: 10.1002/prot.26816
Institution: All Russia Research Institute of Automatics
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4408070177