CD47Binder: Identify CD47 Binding Peptides by Combining Next-Generation Phage Display Data and Multiple Peptide...

Area of Science:

• Immunology
• Computational Biology
• Biotechnology

Background:

• The CD47/SIRPα pathway is a critical target in tumor immunity, emerging after PD-1/PD-L1.
• Current monoclonal antibody therapies for CD47/SIRPα face limitations.
• Peptide-based therapeutics offer potential alternatives with distinct advantages.

Purpose of the Study:

• To develop a predictive computational model for identifying CD47 binding peptides.
• To overcome limitations of existing CD47-targeting antibody therapies.
• To create an accessible bioinformatics tool for CD47 peptide prediction.

Main Methods:

• Utilized next-generation phage display (NGPD) for screening CD47 binding peptides.
• Employed ten traditional machine learning methods and three deep learning methods for model development.
• Integrated multiple peptide descriptors and a support vector machine for the final predictive model.

Main Results:

• The integrated predictive model achieved high performance metrics: 0.755 specificity, 0.764 accuracy, and 0.772 sensitivity via five-fold cross-validation.
• Successfully screened CD47 binding peptides using NGPD biopanning.
• Developed an online bioinformatics tool, CD47Binder, for predicting CD47 binding peptides.

Conclusions:

• The developed integrated model effectively predicts CD47 binding peptides.
• This approach offers a promising strategy for developing novel CD47-targeting cancer immunotherapies.
• The CD47Binder tool provides a valuable resource for researchers in the field.