LEADS-PEP: A Benchmark Data Set for Assessment of Peptide Docking Performance

Area of Science:

• Computational Chemistry
• Drug Discovery
• Structural Biology

Background:

• Peptide-based therapeutics are gaining interest, increasing the need for computational tools like peptide docking.
• Existing peptide docking tools require independent benchmarks for performance assessment and development.

Purpose of the Study:

• To introduce the LEADS-PEP benchmark dataset for evaluating peptide docking performance.
• To assess the suitability of small molecule docking programs for peptide structure prediction.

Main Methods:

• Developed the LEADS-PEP dataset comprising 53 protein-peptide complexes (3-12 residues).
• Evaluated multiple small molecule docking programs against the LEADS-PEP dataset.
• Utilized scoring functions (ChemPLP, ChemScore, ASP) for pose rescoring.

Main Results:

• LEADS-PEP dataset is publicly available at www.leads-x.org.
• Most small molecule docking programs could model small peptides.
• Surflex-Dock and AutoDock Vina performed well for peptides >5 residues.
• Rescoring improved the accuracy of near-native peptide conformations.

Conclusions:

• Small molecule docking programs offer a viable and efficient alternative to specialized peptide docking tools.
• The LEADS-PEP dataset facilitates the development and validation of peptide docking methodologies.