HAMdetector: a Bayesian regression model that integrates information to detect HLA-associated mutations.

Area of Science:

• Immunology
• Virology
• Computational Biology

Background:

• Adaptive immunity relies on human leukocyte antigen (HLA) presenting viral epitopes to CD8+ cytotoxic T-lymphocytes (CTLs).
• Viruses like HIV and HBV evolve immune escape mutations (HAMs) to evade HLA/CTL recognition, posing challenges for treatment and vaccine development.
• Detecting HAMs is vital for understanding viral evolution and designing effective antivirals, but sparse and noisy HLA data complicate identification.

Purpose of the Study:

• To develop a robust computational model for detecting HLA-associated mutations (HAMs) in viral pathogens.
• To integrate diverse data sources, including epitope predictions and phylogenetic information, within a unified Bayesian framework.
• To provide a tool that yields interpretable quantitative information for identifying potential HAM candidates.

Main Methods:

• A novel Bayesian regression model incorporating a sparsity-inducing prior was developed.
• The model integrates epitope prediction algorithms and assesses phylogenetic biases.
• Performance was evaluated using datasets from HBV, HDV, and HIV infections.

Main Results:

• The Bayesian model successfully predicted experimentally confirmed HAMs with high posterior probabilities.
• The developed model demonstrated strong performance compared to existing state-of-the-art methods.
• The approach provides easily interpretable quantitative insights into potential HAMs.

Conclusions:

• The new Bayesian model offers an effective approach for identifying viral immune escape mutations.
• This method aids in understanding viral evolution and developing targeted antiviral therapies and vaccines.
• The software is publicly available, facilitating further research and application.