Prediction and Modeling of Protein-Protein Interactions Using "Spotted" Peptides with a Template-Based Approach

Area of Science:

• Computational biology
• Biochemistry
• Bioinformatics

Background:

• Protein-peptide interactions (PpIs) are crucial for cellular processes and are part of the larger protein-protein interaction (PPI) network.
• Experimental methods for identifying PpIs and PPIs are often time-consuming and expensive.
• In silico prediction offers a viable and cost-effective alternative for studying these interactions.

Purpose of the Study:

• To introduce PepThreader, a novel algorithm and freely available tool for predicting and analyzing protein-peptide interactions (PpIs) and protein-protein interactions (PPIs).
• To provide a computational method that can identify potential binding peptides within protein sequences and model their structure within a target receptor.

Main Methods:

• PepThreader utilizes a threading algorithm that maps protein fragments or peptide libraries onto a template peptide complexed with a protein target.
• The algorithm employs a two-stage scoring system: initial ranking based on sequence similarity, followed by re-ranking using structure-based scoring functions.
• The tool was benchmarked on a dataset of 292 experimentally determined protein-peptide complexes.

Main Results:

• PepThreader achieved an accuracy of 80% when considering the top 25 predicted hits.
• The tool's performance is comparable to existing state-of-the-art methods for PPI and PpI modeling.
• PepThreader uniquely identifies binding peptides within full-length sequences and models their structure within the receptor.

Conclusions:

• PepThreader provides a reliable and efficient in silico method for predicting protein-peptide interactions.
• The tool complements existing experimental techniques, aiding in the identification and characterization of PpIs.
• Its ability to simultaneously spot and model binding peptides offers unique advantages in PPI research.