PreAIP: Computational Prediction of Anti-inflammatory Peptides by Integrating Multiple Complementary Features

Area of Science:

• Biochemistry
• Bioinformatics
• Immunology

Background:

• Therapeutic peptides are considered for inflammatory and autoimmune diseases.
• Experimental identification of anti-inflammatory peptides (AIPs) is costly and time-consuming.
• Development of in silico tools is needed to predict AIPs efficiently.

Purpose of the Study:

• To develop an accurate computational predictor for identifying anti-inflammatory peptides.
• To integrate diverse features for improved prediction accuracy.
• To provide a freely available tool for research.

Main Methods:

• Systematic investigation of primary sequence, evolutionary, and structural features.
• Utilized a random forest classifier for model development.
• Evaluated model performance using 10-fold cross-validation and an independent test dataset.

Main Results:

• The PreAIP model achieved an AUC of 0.833 on the training dataset.
• PreAIP demonstrated superior performance on the test dataset with an AUC of 0.840.
• The developed predictor outperformed existing methods in identifying AIPs.

Conclusions:

• PreAIP is an accurate and effective tool for predicting anti-inflammatory peptides.
• This predictor can accelerate the development of AIPs therapeutics.
• The tool supports advancements in biomedical research for inflammatory diseases.