Is it time to change the reference genome?

Abstract

The use of the human reference genome has shaped methods and data across modern genomics. This has offered many benefits while creating a few constraints. In the following opinion, we outline the history, properties, and pitfalls of the current human reference genome. In a few illustrative analyses, we focus on its use for variant-calling, highlighting its nearness to a 'type specimen'. We suggest that switching to a consensus reference would offer important advantages over the continued use of the current reference with few disadvantages.

Fig. 1

The reference genome is a type specimen. a Cumulative distributions of variants in the reference genome and those in personal/individual genomes. If we collapse the diploid whole genomes genotyped in the 1000 Genomes Project into haploid genomes, we can observe just how similar the reference is to an individual genome. First, taking population allele frequencies from a random sample of 100 individual genomes, we generated new haploid ‘reference’ sequences. We replaced the alleles of the reference genome with the personal homozygous variant, and a randomly chosen heterozygous allele. For simplicity, all calculations were performed against the autosomal chromosomes of the GRCh37 assembly and include only single nucleotide bi-allelic variants (i.e., only two alleles per single nucleotide polymorphism (SNP)). b Cumulative distributions of allele frequencies for variants called in 100 randomly chosen personal genomes, computed against the reference genome. Here, the presence of a variant with respect to the reference is quite likely to mean that the reference itself has the ‘variant’ with respect to any default expectation, particularly if the variant is homozygous

Fig. 2

How consensus alleles improve the interpretability of the reference. a To build a consensus genome, we replaced minor alleles within the current reference with their major alleles (allele frequency (AF) > 0.5) across all bi-allelic SNPs. b Cumulative distributions of variants in the consensus genome (red line) and the current reference (blue line). c Cumulative distributions of AFs for variants in 100 randomly chosen personal genomes, computed against a consensus genome. d Distribution of the number of homozygous single nucleotide variants (SNVs) in 2504 personal genomes, computed against the reference, against an all-human consensus, the mean of the super-population consensuses and the mean of the population consensuses. The consensus reference for each of the five super-populations leads to an additional reduction in the number of homozygous variants in the personal genomes for each super-population (dark red curve). Further breakdown into 26 representative populations does not dramatically reduce the number of homozygous variants (dashed red line). Super-populations are defined broadly as: AFR African, AMR admixed American, EAS East Asian, EUR European, SAS South Asian

Fig. 3

How-to reference. For future or new populations, sequencing is followed by building the consensus sequence from those genomes. Any new genomes will only adjust and improve on the current consensus on the basis of a change in allele frequencies. Finally, the reference can be replicated and diversified into other population-specific references