Predicted and Experimental Peptide Binding Information (PEPBI) Database: A Paired Database of Predicted and Experimental Protein Peptide Binding Information
1. The PEPBI database is a groundbreaking resource for researchers in computational biology, offering a comprehensive collection of 329 predicted peptide-protein complexes, each paired with experimentally determined thermodynamic data. This unique combination of structural and thermodynamic information is expected to significantly aid the development of computational methods for peptide design.
2. The database includes detailed thermodynamic measurements such as changes in Gibbs free energy (ΔG), enthalpy (ΔH), and entropy (ΔS) for each complex. These data are crucial for understanding the binding properties of peptides to proteins, which is essential for applications in sensing, diagnostics, and therapeutics.
3. PEPBI stands out for its stringent selection criteria, ensuring high-quality data. Complexes were selected based on criteria such as peptide length (5–20 residues), structure resolution (≤2.0 Å), and minimal sequence identity between complexes. This ensures a diverse and accurate dataset for computational modeling.
4. The database also includes 40 properties calculated using Rosetta’s Interface Analyzer for each complex. These properties provide a detailed characterization of the binding interactions, which can be invaluable for training machine learning models and improving the accuracy of computational predictions.
5. The development of PEPBI involved a rigorous five-step process, including literature review, computational prediction, and energetic minimization. This meticulous approach ensures that the database is both comprehensive and reliable, making it a valuable tool for researchers in the field.
6. The PEPBI database is freely available and includes a variety of resources such as an Excel spreadsheet, Python script, PDB formatted files, and visual representations of the complexes. This accessibility makes it easy for researchers to integrate the data into their workflows and develop new computational methods.
📜Paper: https://t.co/ahIyzdSJPP
#ComputationalBiology #ProteinPeptideBinding #Database #Thermodynamics #PeptideDesign #StructuralBiology
Check out the latest paper from my PhD work!
In it we determined features of Antibody-Antigen Interfaces that are critical to binding. We applied those interface features using a random forrest classifier and were able to significantly reduce false positive binding poses.