Using Gaussian process with test rejection to detect T-cell epitopes in pathogen genomes.

Area of Science:

• Computational biology
• Vaccine development
• Immunoinformatics

Background:

• Developing peptide-based vaccines faces challenges in identifying potent immunogenic elements from vast pathogen genomes.
• Existing computational methods for predicting epitopes often yield numerous false positives, complicating experimental validation.

Purpose of the Study:

• To develop and evaluate a novel computational method for enhancing the accuracy of T-cell epitope prediction.
• To reduce false positives in epitope prediction, thereby improving the efficiency of vaccine candidate selection.

Main Methods:

• Utilized Gaussian process regression to estimate prediction uncertainty for T-cell epitopes.
• Developed a test reject method based on these uncertainty estimates to filter false positives.
• Compared the performance of Gaussian process regression against state-of-the-art methods using cross-validation on 11 benchmark datasets.

Main Results:

• Gaussian process regression demonstrated comparable accuracy to top-performing algorithms in epitope prediction.
• The proposed test reject method significantly reduced false positives without compromising prediction sensitivity.
• Application to the Vaccinia virus genome successfully identified true T-cell epitopes.

Conclusions:

• Gaussian process regression combined with a test reject strategy offers an effective approach for predicting T-cell epitopes.
• This method is particularly valuable for analyzing large and diverse pathogen genomes where minimizing false positives is critical for vaccine design.