Repository for code, summary statistic data and trained HLA models
Publication:
Ritari J, Hyvärinen K, Clancy J, FinnGen, Partanen J, Koskela S. Increasing accuracy of HLA imputation by a population-specific reference panel in a Finngen biobank cohort. NAR Genomics and Bioinformatics, Volume 2, Issue 2, June 2020, lqaa030, https://doi.org/10.1093/nargab/lqaa030
results.R R code that generates the plots for figures 2-5
functions.R helper functions for e.g. plotting
HLA_imputation.R example of running imputation with Finnish reference panel
FG_HLA_impute.R R code for running FinnGen HLA imputation and result BGEN file production
Trained HLA imputation models for hg19 and hg38 human genome builds. The models for DRB3-5 genes are in hg38 build only. The 'ng' in DRB3-5 is not an allele as such, but indicates that the gene is missing.
Contains imputation error rates, autoimmune association summaries and HLA allele frequency information
MHC region SNPs were first converted to plink format (.bed, .bim, .fam) from VCF genotypes using tabix and plink, and subequently processed with R. HLA alleles for HLA-A, -B, -C, -DPB1, -DQA1, -DQB1, -DRB1, and -DRB3-5 genes were imputed with the HIBAG R library using a Finnish reference panel. Imputation posterior probabilities (pp) for each imputed allele were extracted from the HIBAG output by summing the pp values from all imputed HLA genotypes of which a given allele was a part of. The pp values were imported to plink2 using --import-dosage command and converted to BGEN format with --export bgen-1.2 bits=16 ref-first.