همایش ملی بیوانفورماتیک ایران

صفحه اصلی / 4th international edition and 13th Iranian Conference on Bioinformatics

Harnessing Deep Learning for Epitope Prediction in Immunoinformatics: A Novel Framework for Vaccine Design

نویسندگان :

Ali Rahmati hagh bonab¹ Hannaneh sadat Jalilzadeh²

1- دانشگاه علوم پزشکی شهید بهشتی 2- دانشگاه علوم پزشکی شهید بهشتی

کلمات کلیدی :

Deep Learning،Immunoinformatics،Neural Networks،Epitope Prediction

چکیده :

Background: Immunoinformatics and deep learning have emerged as synergistic tools in the identification of epitopes for vaccine development. Traditional computational approaches often fail to capture the complex, nonlinear relationships inherent in immune recognition. Materials and Methods: In this study, we developed and implemented a deep learning framework tailored for epitope prediction. Our model was trained on diverse datasets comprising experimentally validated T-cell and B-cell epitopes, utilizing recurrent neural networks (RNN) and convolutional neural networks (CNN). Advanced feature extraction methods were applied to analyze peptide sequences and HLA-binding affinities. The performance of the model was validated against benchmark datasets and compared with existing epitope prediction tools using precision, recall, and F1-score metrics. Results: Our deep learning framework achieved a substantial improvement in epitope prediction accuracy, with an F1-score of 0.92, outperforming state-of-the-art tools. The model demonstrated robust generalizability across multiple antigen types and successfully identified novel epitopes for SARS-CoV-2 and other pathogens. Conclusion: The integration of deep learning with immunoinformatics presents a powerful approach for epitope prediction, accelerating vaccine design and advancing our understanding of host-pathogen interactions. Our results underline the potential of data-driven methods to complement experimental immunology.