PPI-Affinity

PPI-Affinity predicts binding affinities and generates ranked mutants for protein–protein and protein–peptide complexes to support hit identification and lead optimization in peptide-based drug design.

Key Features:

• Support Vector Machine (SVM) predictors: Leverages SVM predictors to model interactions characteristic of peptide binders.
• Complex types handled: Screens datasets containing protein–protein (PPi) and protein–peptide complexes.
• Ranked mutant generation: Generates ranked mutants for a given structure to optimize binding affinities.
• Benchmark evaluation: SVM models were evaluated using four benchmark datasets encompassing PPi and protein–peptide binding affinity data.
• Biological validation: Performance was validated on biological systems, including mutants of EPI-X4 (an endogenous peptide inhibitor of CXCR4) and complexes involving serine proteases HTRA1 and HTRA3 with peptides.
• Peptide-focused modeling: Addresses limitations of models trained on small molecules by using predictors tailored to peptide interactions.
• Application to screening and optimization: Targets virtual screening for hit identification and lead optimization in peptide-based drug design.

Scientific Applications:

• Binding affinity prediction: Predicts binding affinities for protein–protein and protein–peptide complexes.
• Virtual screening: Supports hit identification in peptide-based virtual screening workflows.
• Lead optimization: Generates ranked mutants to guide lead optimization of peptide binders.
• Experimental validation support: Applied to validation studies including EPI-X4 mutants and HTRA1/HTRA3–peptide complexes.

Methodology:

Leverages support vector machine (SVM) predictors, screens datasets of protein–protein (PPi) and protein–peptide complexes, generates ranked mutants for given structures, and evaluates SVM models using four benchmark datasets.