The Great Genotyper: a graph-based method for population genotyping of small and structural variants.

Background: Long-read sequencing (LRS) enables high-quality structural variant (SV) discovery. SV genotypers utilize these precise call sets to improve the recall and precision of genotyping in short-read sequencing (SRS) samples. With the extensive growth in publicly available SRS datasets, it is now possible to calculate accurate population allele frequencies of SVs. However, reprocessing hundreds of terabytes of raw SRS data to genotype new variants is impractical for population-scale studies, a computational challenge known as the N+1 problem (i.e., the challenge of re-genotyping an entire cohort for one additional variant). Overcoming this computational bottleneck is essential for analyzing new SVs from the growing number of pangenomes, public genomic databases, and pathogenic variant discovery studies.

Results: We propose the Great Genotyper, a population-scale genotyping workflow to address the N+1 problem. Applied to a human dataset, the workflow begins by preprocessing 4.2k short-read samples of a total of 183 TB raw data to create an 867-GB Counting Colored de Bruijn Graph (CCDG). The Great Genotyper uses this CCDG to genotype a list of phased or unphased variants, leveraging the CCDG population information to increase both precision and recall. The Great Genotyper offers the same accuracy as the state-of-the-art genotypers while achieving unprecedented performance. It took about 100 hours to genotype 4.5M variants across the 4.2k samples and calculate their population allele frequencies using 1 server with 32 cores and 145 GB of memory. The Great Genotyper opens the door to new ways to study SVs. For example, using the premade index, we demonstrate the Great Genotyper's application in finding pathogenic variants by calculating accurate allele frequency for novel SVs. Also, we used it to create a 4k reference panel by genotyping variants from the Human Pangenome Reference Consortium (HPRC). The new reference panel allows for SV imputation from genotyping microarrays. Moreover, we genotype the human GWAS Catalog and merge its variants with the 4k reference panel. We show 6,253 events of high linkage between the HPRC's SVs and nearby GWAS single-nucleotide polymorphisms, which can help in interpreting the effect of these SVs on gene functions. This analysis uncovers the detailed haplotype structure of the human fibrinogen locus and revives the pathogenic association of a 28-bp insertion in the FGA gene with thromboembolic disorders.

Conclusion: The Great Genotyper solves the N+1 problem for population-scale genotyping of small and structural variants, offering both high accuracy and efficiency. Its ability to rapidly re-genotype large cohorts paves the road for several new studies of SVs.