A chemogenomics view on protein-ligand spaces

Area of Science:

• Computational chemistry and structural biology.
• Bioinformatics and cheminformatics.
• Drug discovery and development.

Background:

• Chemogenomics is an interdisciplinary field merging chemistry, biology, and informatics for drug discovery.
• It aims to understand molecular recognition between ligands and drug targets.
• Existing studies often analyze protein and ligand spaces separately, necessitating methods to explore their joint space.

Purpose of the Study:

• To develop and present a method for visualizing protein-ligand spaces from a chemogenomics perspective.
• To integrate both ligand and protein features for a comprehensive analysis.
• To apply this visualization approach to structural and drug-target interaction datasets.

Main Methods:

• Utilized two chemogenomics datasets: Protein Data Bank (PDB) interactions and DrugBank approved drugs/targets.
• Computed sequence-based descriptors for proteins and 0-2D descriptors for ligands.
• Applied Principal Component Analysis (PCA) for multidimensional data analysis and nearest neighbor methods to assess dataset overlap.

Main Results:

• Generated global models of protein-ligand space using PCA.
• Quantified overlap between different protein-ligand interaction datasets.
• Successfully visualized and compared structural protein-ligand space and approved drug-target space.

Conclusions:

• The presented approach effectively visualizes protein-ligand spaces by considering both molecular components.
• This method is applicable to diverse protein-ligand interaction datasets.
• The visualization can aid in identifying cross-interaction complexes within the protein-ligand space.