A vision of how low-coverage sequence data should contribute to genetic evaluation in the future

Low-coverage sequencing refers to sequencing DNA of individuals to a low depth of coverage (e.g., 0.5X) and imputing that sequence to a genomic sequence based on reference haplotypes from individuals sequenced to a high depth of coverage (e.g., ≥10X). It has been proposed as an alternative to genotyping by Single-nucleotide polymorphisms (SNP) arrays. At least one commercial product based on it is available for agricultural species. Concerns limiting adoption in its current form are: 1) the cost of storing the huge volume of data it generates and 2) whether that additional data will result in improved accuracy of genetic evaluation. This work envisions future implementation of low-coverage sequencing to reduce storage costs and enhance genetic evaluations by leveraging the additional information in the full sequence of the pangenome to account for more genetic variation. We propose addressing the storage issue by representing genomic sequence of an individual in a pair of haplotype arrays with each element pointing to an enumerated haplotype of the sequence within one of approximately 50,000 defined genome segments. Assuming 60 million genomic variants, the infrastructure required to translate the identifier of any enumerated haplotype into its genomic sequence would require less than 10 gigabytes of binary storage. Each haplotype array element would require 2 bytes, so the marginal binary storage required to represent the genomic sequence of an individual would be about 200 kilobytes (KB), similar to the genotypes from a SNP array with 200,000 markers. This assumes no pedigree and no ambiguity of the imputation, though the latter is unrealistic. Strategies to minimize, and when necessary, to manage and efficiently represent ambiguity are proposed. The genomic sequence of an individual could be stored in about 1 KB (binary) if both parents have unambiguous sequences stored as described above. The proposed system for representing the pangenome includes algorithms for read mapping and imputation intended to leverage all known genetic variation in the target population. It is also designed to use sequencing reads generated for imputing the genomic sequence of new individuals to identify unrecognized mutations, crossovers, and structural variants, thus continuously improving the genome representation, especially if widespread use of low-coverage sequencing in livestock industries is realized. This could make improved genetic merit and management of livestock feasible without computational burden.