Genotyping of classical HLA alleles is an essential tool in the analysis of diseases and adverse drug reactions with associations mapping to the major histocompatibility complex (MHC). However, deriving high-resolution HLA types subsequent to whole-genome SNP typing or sequencing is often cost prohibitive for large samples. An alternative approach takes advantage of the extended haplotype structure within the MHC to predict HLA alleles using dense SNP genotypes, such as those available from genome-wide SNP panels. Current methods for HLA imputation are difficult to apply or may require the user to have access to large training data sets with SNP and HLA types. We propose HIBAG, HLA Imputation using attribute BAGging, that makes predictions by averaging HLA type posterior probabilities over an ensemble of classifiers built on bootstrap samples. We assess the performance of HIBAG using our study data (n = 2, 668 subjects of European ancestry) as a training set and HLA data from the British 1958 birth cohort study (n ≈ 1, 000 subjects) as independent validation samples. Prediction accuracies for HLA–A, B, C, DRB1 and DQB1 range from 92.2% to 98.1% using a set of SNP markers common to the Illumina 1M Duo, OmniQuad, OmniExpress, 660K and 550K platforms. HIBAG performed well compared to the other two leading methods HLA*IMP and BEAGLE. This method is implemented in a freely-available HIBAG R package that includes pre-fit classifiers for European, Asian, Hispanic and African ancestries, providing a readily available imputation approach without the need to have access to large training datasets.

Statistical imputation of classical HLA alleles in case-control studies has become established as a valuable tool for identifying and fine-mapping signals of disease association in the MHC. Imputation into diverse populations has, however, remained challenging, mainly because of the additional haplotypic heterogeneity introduced by combining reference panels of different sources. We present an HLA type imputation model, HLA*IMP:02, designed to operate on a multi-population reference panel. HLA*IMP:02 is based on a graphical representation of haplotype structure. We present a probabilistic algorithm to build such models for the HLA region, accommodating genotyping error, haplotypic heterogeneity and the need for maximum accuracy at the HLA loci, generalizing the work of Browning and Browning (2007) and Ron et al. (1998). HLA*IMP:02 achieves an average 4-digit imputation accuracy on diverse European panels of 97% (call rate 97%). On non-European samples, 2-digit performance is over 90% for most loci and ethnicities where data available. HLA*IMP:02 supports imputation of HLA-DPB1 and HLA-DRB3-5, is highly tolerant of missing data in the imputation panel and works on standard genotype data from popular genotyping chips. It is publicly available in source code and as a user-friendly web service framework.

While much is known about human genetic variation, such information is typically ignored in assembling novel genomes. Instead, reads are mapped to a single reference, which can lead to poor characterization of regions of high sequence or structural diversity. We introduce a population reference graph, which combines multiple reference sequences and catalogues of variation. The genomes of novel samples are reconstructed as paths through the graph using an efficient hidden Markov model, allowing for recombination between different haplotypes and additional variants. By applying the method to the 4.5Mb extended MHC region on human chromosome 6, combining eight assembled haplotypes, sequences of known classical HLA alleles and 87,640 SNP variants from the 1000 Genomes Project, we demonstrate, using simulations, SNP genotyping, short-read and long-read data, how the method improves the accuracy of genome inference and reveals regions where the current set of reference sequences is substantially incomplete.