Abstract
AbstractDeepVASP-S is a computational tool that leverages the capabilities of convolutional neural networks (CNNs) to analyze steric aspects of protein-ligand interactions and predict amino acid contributions to binding specificity. This tool combines structural bioinformatics with machine learning to address the complex problem of understanding how specific amino acids contribute to the specificity of binding. Here, we use this tool to predict subclasses for Enolase and Serine Protease according to their binding specificity, and explain it through the underlying steric mechanism. The strength of DeepVASP-S lies in its ability to identify and highlight these specific regions, providing researchers with insights into the molecular determinants of protein function and interaction. Such information is extremely valuable for the field of drug design, as it enables the creation of more targeted and effective therapeutics with minimized side effects. The approach taken by DeepVASP-S represents a significant step forward in computational biochemistry, merging high-resolution 3D structural data with the predictive power of machine learning to unlock a deeper understanding of protein-ligand interactions.
Publisher
Cold Spring Harbor Laboratory