Benchmark Sets for Binding Hot Spot Identification in Fragment-Based Ligand Discovery

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Binding hot spots are protein regions with high affinity for small molecules, contributing significantly to binding energy.
• Identifying these hot spots is key for efficient fragment-based ligand discovery.
• Existing benchmark sets have limitations for evaluating hot spot prediction methods.

Purpose of the Study:

• To present novel benchmark sets for assessing computational methods aimed at identifying protein binding hot spots.
• To facilitate the evaluation of tools used in fragment-based ligand discovery.
• To establish a baseline for testing prediction methods using FTMap.

Main Methods:

• Creation of benchmark sets including protein structures with bound fragments and corresponding unbound structures.
• Inclusion of a set from Astex Pharmaceuticals focusing on fragment binding validation.
• Utilizing FTMap, a computational tool analogous to fragment screening, to test the benchmark sets.

Main Results:

• The study presents four distinct benchmark sets for evaluating hot spot identification methods.
• FTMap was used to establish a performance baseline across these sets.
• Differences and characteristics of the benchmark sets were analyzed.

Conclusions:

• The developed benchmark sets provide a valuable resource for testing and validating computational hot spot prediction tools.
• These sets are particularly relevant for fragment-based drug discovery efforts.
• The study highlights the importance of both bound and unbound structures in benchmark set design.