The druggable genome: Twenty years later

Area of Science:

• Drug discovery and bioinformatics
• Genomics and proteomics
• Computational biology and artificial intelligence

Background:

• The concept of the druggable genome has been explored for two decades, with various methods developed to assess target druggability.
• Resources like Open Targets collate target-disease associations, and the Protein Data Bank in Europe (PDBe) links residue-level annotations to protein structures.

Purpose of the Study:

• To integrate diverse biological data, from gene to protein residue levels, into a unified knowledge graph.
• To develop scalable, automated workflows for hotspot-based druggability assessments of protein structures.
• To leverage graph-based AI for navigating complex biological data and identifying future drug targets.

Main Methods:

• Development of scalable, automated workflows for processing protein structures and assessing druggability.
• Integration of data from gene-level annotations to per-residue protein structure information.
• Utilizing graph-based artificial intelligence (AI) methods for knowledge graph navigation.

Main Results:

• Established automated workflows for hotspot-based druggability assessments across numerous targets.
• Created a framework for integrating multi-level biological data into a comprehensive knowledge graph.
• Demonstrated the potential for proteome-scale analysis, facilitated by advancements like AlphaFold 2.

Conclusions:

• Integrating gene-level to residue-level data into a knowledge graph creates a powerful resource for drug discovery.
• Automated workflows and graph-based AI are essential for managing and extracting insights from complex biological data.
• This integrated approach promises to significantly advance the identification of novel drug targets.