UPDATE: This paper has now been published in Genes.
A new preprint from scientists at Uppsala University’s SciLifeLab reports the de novo genome sequencing and assembly of two Swedish individuals using PacBio SMRT Sequencing. By comparing the Swedish genomes to the human reference (GRCh38), the team found a substantial amount of novel sequence which is not present in the reference – along with over 17,000 structural variants. Further comparison of the Swedish genomes to other population-specific reference genome assemblies – including a Korean and a Chinese genome – identified novel sequences that appear to be population-specific as well as several megabases that seem to be more universal in the human genome.
“De novo assembly of two Swedish genomes reveals missing segments from the human GRCh38 reference and improves variant calling of population-scale sequencing data” comes from lead author Adam Ameur, senior author Ulf Gyllensten, and collaborators. They aimed to produce higher-quality genome assemblies than are possible with short-read sequencing technologies. “PacBio’s single-molecule real-time (SMRT) sequencing technology has proven to be an excellent method for de novo genome assembly,” they write. “The human de novo assemblies available based on long-read data … indicate that each personal genome contains a significant amount of ‘dark matter’ of structural variation that is not detected by short-read WGS.”
The scientists generated about 78-fold SMRT Sequencing coverage of each genome (one male and one female), followed by optical mapping for scaffolding. The PacBio-only assemblies were highly contiguous: authors report that each one “contained about 3,000 primary contigs and an additional 4,000 alternative contigs originating from regions with high heterozygosity.” For primary contigs, N50 values were 9.5 Mb and 8.5 Mb.
They then compared the assemblies to each other, to GRCh38, and to other de novo assemblies recently produced with SMRT Sequencing. More sequence matched between the two Swedish genomes than with GRCh38, “suggesting that the [human] reference does not contain all sequences present in these Swedish individuals,” the scientists report, citing about 10 Mb absent from the reference. Of the novel sequences discovered, about 6 Mb aligned to a Chinese genome assembly, indicating that much of the data missing from GRCh38 is not specific to the Swedish population. Novel sequences had the typical hallmarks that make detection difficult for short-read sequencers: they “are highly repetitive, have elevated GC-content and are primarily located in centromeric or telomeric regions,” according to the preprint.
“Inclusion of these novel sequences into the GRCh38 reference radically improves the alignment and variant calling of whole-genome sequencing data at several genomic loci,” the scientists add. By re-analyzing 200 samples from a short-read-based Swedish population study, the team found more than 75,000 putative novel SNVs in each person and removed 10,000 SNV calls per person that had been false positives.
“The benefits of an improved reference are likely to be even stronger for other, non-European, population groups that were poorly represented in the original assembly of GRCh38. Despite all efforts to refine the human genome since its original release in 2001, our results indicate that substantial improvements could still be made … by de novo assembly of representative human genomes from different populations,” the scientists conclude.