CD47Binder

CD47Binder identifies peptides that bind the CD47 protein by combining next-generation phage display (NGPD) data with machine learning to predict CD47-binding peptides for tumor-immunity research.

Key Features:

• Next-Generation Phage Display (NGPD): High-throughput NGPD biopanning of peptide libraries is used to generate candidate CD47-binding sequences.
• Machine Learning Integration: Ten traditional machine learning algorithms and three deep learning models are applied using multiple peptide descriptors to refine binding predictions.
• Predictive Model: An integrated support vector machine (SVM) model is developed and validated by five-fold cross-validation with reported specificity 0.755, accuracy 0.764, and sensitivity 0.772.

Scientific Applications:

• Tumor immunity research: Identification of CD47-binding peptides to investigate CD47-mediated immune evasion mechanisms in oncology studies.
• Therapeutic candidate discovery: Prioritization of peptide candidates for development of anti-tumor therapies targeting the CD47/SIRPα pathway.
• Alternative to monoclonal antibodies: Generation of peptide leads that may address limitations associated with existing monoclonal antibody treatments targeting CD47.

Methodology:

Application of ten traditional machine learning algorithms and three deep learning models using multiple peptide descriptors, integration into an SVM predictive model, and validation by five-fold cross-validation (specificity 0.755, accuracy 0.764, sensitivity 0.772).