SEP-AlgPro: An efficient allergen prediction tool utilizing traditional machine learning and deep learning techniques...

Area of Science:

• Bioinformatics
• Computational Biology
• Immunology

Background:

• Allergies involve hypersensitive reactions to typically harmless substances.
• Current computational allergen identification methods lack accuracy and efficiency due to limited features and datasets.

Purpose of the Study:

• To develop an accurate computational method for identifying allergen proteins from sequence information.
• To compare the effectiveness of traditional protein features versus protein language model features for allergen prediction.

Main Methods:

• Analyzed 10 conventional protein features and 14 protein language model features.
• Utilized 15 different classifiers, optimizing with top features and classifiers.
• Employed a deep neural network integrating predictions from optimized baseline models.

Main Results:

• Protein language model features demonstrated superior discriminative power over traditional features.
• The optimized model aggregated predictions from baseline models for enhanced accuracy.
• SEP-AlgPro significantly outperformed existing state-of-the-art allergen predictors.

Conclusions:

• SEP-AlgPro provides a highly accurate and efficient method for allergen protein identification.
• Protein language models are crucial for improving allergen prediction.
• A freely available web server enhances accessibility for researchers and clinicians.