A multi-ethnic reference panel to impute HLA classical and non-classical class I alleles in admixed samples: Testing...

Area of Science:

• Immunogenetics
• Bioinformatics
• Population Genetics

Background:

• The human leukocyte antigen (HLA) region is critical for immune responses and disease susceptibility.
• Genome-wide association studies (GWAS) identify single nucleotide polymorphisms (SNPs) but don't fully capture HLA allele relevance.
• Accurate HLA imputation relies on large, diverse reference panels, particularly for admixed populations.

Purpose of the Study:

• To develop and validate novel, multi-ethnic reference panels for HLA imputation using whole-genome sequencing data.
• To assess the performance of different reference panel compositions, including multi-ethnic and population-specific panels.
• To expand imputation capabilities to non-classical HLA genes (MICA, MICB, HLA-H).

Main Methods:

• Utilized bioinformatics approaches to call SNPs and HLA alleles from 1000 Genomes (1KG) and Brazilian SABE datasets (30X WGS data).
• Created three reference panels: 1KG (n=2504), SABE (n=1171), and a combined full model (n=3675) using HIBAG.
• Performed extensive cross-validation and validated performance on an independent Brazilian dataset, comparing against the Michigan Imputation Server.

Main Results:

• The multi-ethnic 1KG reference panel outperformed the Brazilian-only panel.
• The combined full model reference panel yielded the best imputation results.
• Developed novel reference panels for non-classical HLA genes (MICA, MICB, HLA-H).
• The developed panels demonstrated superior performance over the Michigan Imputation Server, especially for HLA-B allele imputation in Brazilians.

Conclusions:

• Multi-ethnic reference panels are crucial for accurate HLA imputation across diverse populations.
• Adapting reference panels to match the genetic diversity of the target population is essential for optimal imputation.
• The developed reference panels significantly improve HLA imputation accuracy, particularly for underrepresented admixed populations.