Search

Reference genomes provide mapping targets and coordinate systems but introduce biases when samples under study diverge sufficiently from them. Pangenome references seek to address this by storing a representative set of diverse haplotypes and their alignment, usually as a graph. Alternate alleles determined by variant callers can be used to construct pangenome graphs, but thanks to advances in long-read sequencing, high-quality phased assemblies are becoming widely available. Constructing a pangenome graph directly from assemblies, as opposed to variant calls, leverages the graph’s ability to consistently represent variation at different scales and reduces biases introduced by reference-based variant calls. Pangenome construction in this way is equivalent to multiple genome alignment. Here we present the Minigraph-Cactus pangenome pipeline, a method to create pangenomes directly from whole-genome alignments, and demonstrate its ability to scale to 90 human haplotypes from the Human Pangenome Reference Consortium (HPRC). This tool was designed to build graphs containing all forms of genetic variation while still being practical for use with current mapping and genotyping tools. We show that this graph is useful both for studying variation within the input haplotypes, but also as a basis for achieving state of the art performance in short and long read mapping, small variant calling and structural variant genotyping. We further measure the effect of the quality and completeness of reference genomes used for analysis within the pangenomes, and show that using the CHM13 reference from the Telomere-to-Telomere Consortium improves the accuracy of our methods, even after projecting back to GRCh38. We also demonstrate that our method can apply to nonhuman data by showing improved mapping and variant detection sensitivity with aDrosophila melanogasterpangenome.

Methods that use a linear genome reference for genome sequencing data analysis are reference biased. In the field of clinical genetics for rare diseases, a resulting reduction in genotyping accuracy in some regions has likely prevented the resolution of some cases. Pangenome graphs embed population variation into a reference structure. While pangenome graphs have helped to reduce reference mapping bias, further performance improvements are possible. We introduce VG-Pedigree, a pedigree-aware workflow based on the pangenome-mapping tool of Giraffe (Sirén et al. 2021) and the variant-calling toolDeepTrio(Kolesnikov et al. 2021) using a specially-trained model for Giraffe-based alignments. We demonstrate mapping and variant calling improvements in both single-nucleotide variants (SNVs) and insertion and deletion (INDEL) variants over those produced by alignments created using BWA-MEM to a linear-reference and Giraffe mapping to a pangenome graph containing data from the 1000 Genomes Project. We have also adapted and upgraded the deleterious-variant (DV) detecting methods and programs of Gu et al. into a streamlined workflow (Gu et al. 2019). We used these workflows in combination to detect small lists of candidate DVs among 15 family quartets and quintets of the Undiagnosed Diseases Program (UDP). All candidate DVs that were previously diagnosed using the mendelian models covered by the previously published Gu et al. methods were recapitulated by these workflows. The results of these experiments indicate a slightly greater absolute count of DVs are detected in the proband population than in their matched unaffected siblings.

There is increasing recognition that a single, monoploid reference genome is a poor universal reference structure for human genetics, because it represents only a tiny fraction of human variation. Adding this missing variation results in a structure that can be described as a mathematical graph: a genome graph. We demonstrate that, in comparison to the existing reference genome (GRCh38), genome graphs can substantially improve the fractions of reads that map uniquely and perfectly. Furthermore, we show that this fundamental simplification of read mapping transforms the variant calling problem from one in which many non-reference variants must be discovered de-novo to one in which the vast majority of variants are simply re-identified within the graph. Using standard benchmarks as well as a novel reference-free evaluation, we show that a simplistic variant calling procedure on a genome graph can already call variants at least as well as, and in many cases better than, a state-of-the-art method on the linear human reference genome. We anticipate that graph-based references will supplant linear references in humans and in other applications where cohorts of sequenced individuals are available.