Machine Learning-Enhanced T Cell Neoepitope Discovery for Immunotherapy Design.

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• T cell immune responses target specific peptides (T cell epitopes) presented by human leukocyte antigen (HLA) molecules.
• HLA gene polymorphism influences peptide binding and can generate neoepitopes, altering T cell responses.
• Machine learning (ML) algorithms show promise in accurately predicting HLA-peptide binding affinity.

Purpose of the Study:

• To highlight the role of machine learning in predicting T cell epitopes.
• To discuss the challenges and future directions in neoepitope prediction.
• To underscore the value of accurate neoepitope prediction for personalized immunotherapies.

Main Methods:

• Utilizing machine learning algorithms for predicting HLA-peptide binding affinity.
• Leveraging next-generation sequencing data for neoepitope identification.
• Developing integrated pipelines for neoepitope prediction and analysis.

Main Results:

• ML tools can predict HLA-peptide binding affinity with considerable accuracy.
• Polymorphisms in HLA genes can lead to the generation of neoepitopes.
• Accurate neoepitope prediction is crucial for understanding T cell responses.

Conclusions:

• Machine learning has significantly advanced the prediction of neoepitopes from *in silico* data.
• Overcoming challenges in dataset availability and pipeline integration is key.
• Precise neoepitope prediction holds great potential for personalized cancer immunotherapies and other treatments.