ppdbench

ppdbench benchmarks protein-peptide docking predictions by calculating CAPRI parameters to assess the accuracy of docked models relative to native complexes.

Key Features:

• CAPRI parameter calculation: Computes Fraction of Native Contacts (FNAT), Ligand-RMSD (L-RMSD), and Interface-RMSD (I-RMSD) between native and docked complexes.
• Benchmark dataset: Evaluates docking performance across 133 protein-peptide complexes with peptide lengths of 9–15 residues.
• Method comparison: Benchmarks six different docking methods and compares their predictions using CAPRI metrics.
• Blind docking versus re-docking: Distinguishes performance in blind docking and re-docking scenarios and reports separate results.
• Pose ranking analysis: Reports and contrasts average L-RMSD for top poses and best poses to evaluate ranking effectiveness.
• Subset evaluation: Includes analysis on a subset of 40 protein-peptide complexes with peptide lengths up to 5 residues, highlighting method-specific performance differences (e.g., FRODOCK, ZDOCK, AutoDock Vina).
• Performance reporting: Records method-specific numerical outcomes such as average L-RMSD values for comparative assessment.

Scientific Applications:

• Docking validation: Validates and benchmarks protein-peptide docking methods using standardized CAPRI metrics.
• Method selection and improvement: Informs selection of docking algorithms (e.g., FRODOCK, ZDOCK, AutoDock Vina) and highlights the need for improved pose-ranking schemes.
• Research support: Supports studies in drug discovery, molecular biology, and related fields that require assessment of protein-peptide interaction models.

Methodology:

Calculates CAPRI parameters (FNAT, L-RMSD, I-RMSD) between native and docked complexes, computes average L-RMSD for top and best poses, benchmarks six docking methods on 133 protein-peptide complexes (peptide lengths 9–15 residues), and evaluates a subset of 40 complexes (peptide lengths up to 5 residues) including blind docking and re-docking comparisons.