SimPep and OP-AND: A deep learning framework and curated database for predicting osteogenic peptides.

Area of Science:

• Biochemistry
• Computational Biology
• Nutritional Science

Background:

• Bone health is a significant concern for aging populations worldwide.
• Bioactive peptides from dairy products show potential for preventing bone diseases.
• Limited public databases for osteogenic peptides (OPs) hinder computational research.

Purpose of the Study:

• To create a public database of osteogenic peptides (OPs).
• To develop a computational framework for identifying OPs.
• To propose a hypothesis regarding non-osteogenic peptides involved in osteoclast formation.

Main Methods:

• Development of the OP-AND database, a curated collection of OPs.
• Creation of SimPep, a deep learning framework for OP prediction.
• 5-fold cross-validation and evaluation on external datasets.

Main Results:

• The SimPep framework achieved 86.87% accuracy and 76.88% AUC.
• A pipeline for selecting potential OPs for experimental validation was introduced.
• Camel milk alpha s1-casein peptide 'MKLLILTCLVAVALARPKYPLRYPEVF' identified as a top candidate.

Conclusions:

• OP-AND and SimPep provide valuable resources for osteogenic peptide research.
• The developed methods facilitate the discovery of novel OPs for bone health.
• Further experimental validation of identified candidate peptides is warranted.