HAMdetector

HAMdetector identifies Human Leukocyte Antigen (HLA)-associated mutations in viral genomes using a Bayesian regression framework to detect immune-escape variants that affect MHC I presentation to CD8+ cytotoxic T-lymphocytes.

Key Features:

• Bayesian Inference: Uses a Bayesian regression model to integrate multiple information sources for detection of HLA-associated mutations (HAMs).
• Sparsity-Inducing Priors: Applies sparsity-inducing priors to promote sparse solutions and facilitate identification of candidate HAMs with interpretable quantitative metrics.
• Integrative Priors: Incorporates prior knowledge including epitope affinities to MHC I, peptide processing predictions, and phylogenetic background analysis.
• Paired Data Integration: Analyzes paired viral sequence and HLA genotype data to associate specific mutations with HLA alleles.
• Noisy and Sparse Data Handling: Addresses sparsity and noise arising from high HLA polymorphism through its Bayesian framework.
• Improved Performance: Demonstrates improved detection performance compared to existing approaches by leveraging comprehensive data integration.

Scientific Applications:

• Viral evolution and immune escape studies: Identifies HAMs to study selective pressure exerted by CD8+ T-lymphocytes and mechanisms of immune escape.
• Vaccine and antiviral design: Informs development of antiviral strategies and vaccines by pinpointing mutations that enable immune escape, particularly in HIV and HBV.

Methodology:

Bayesian regression integrating paired sequence and HLA data with sparsity-inducing priors and priors derived from epitope–MHC I affinities, peptide processing predictions, and phylogenetic background analysis.