PHG

PHG constructs and queries a pangenome haplotype graph to store haplotypes and variant information and to impute missing genotypes for genome-wide imputation and genomic prediction in breeding programs.

Key Features:

• Pangenome database: Stores all identified haplotypes and variant information across a set of individuals.
• PHG construction at multiple scales: Supports PHG builds comprising different sample sets, exemplified by a 24-individual and a 398-individual PHG.
• Genic-region diversity capture: Captures genetic diversity present within genic regions of the sorghum genome.
• Low-coverage SNP calling: Calls single nucleotide polymorphisms (SNPs) accurately at very low sequencing coverage (reported error rate 5.9% at 0.01x) with performance near that at higher coverage (8x).
• Genotype imputation: Imputes missing genotypes in progeny using parental haplotypes stored within the PHG.
• Cross-platform genotype unification: Unifies genotype calls from different sequencing platforms and methods, including genotyping-by-sequencing (GBS) and rhAmpSeq.
• Efficient storage: Stores sequence and variant data in an efficient format to reduce input sequence requirements and genotyping costs.
• Genomic prediction support: Enables genomic prediction using imputed genotypes with reported mean accuracies between 0.57 and 0.73 for various traits.

Scientific Applications:

• Genomic selection in breeding programs: Applied to the Chibas sorghum breeding program by sequencing 24 founders at low coverage and imputing genotypes in progeny for selection.
• Genotype imputation in progeny: Imputes genotypes in 207 progeny using parental haplotypes stored in the PHG.
• Comparative genotyping: Provides genotype datasets comparable to genotyping-by-sequencing (GBS) and sequencing targeted amplicons (rhAmpSeq) for downstream analyses.
• Cost-effective genotyping for large populations: Reduces required sequencing input to lower genotyping costs and enable larger breeding populations to capture genetic recombination.
• Cross-species pangenome storage: Maintains comprehensive variant information applicable to diverse taxa and any species.

Methodology:

Construct PHG pangenome databases storing haplotypes and variants (examples: 24- and 398-individual PHGs), perform SNP calling from low-coverage sequencing (reported 0.01x and 8x comparisons), and impute missing progeny genotypes using parental haplotypes stored in the PHG, as demonstrated by processing 24 low-coverage founders and imputing 207 progeny.