Ultra-long reads and ultra-long duplications: What nanopore sequencing is revealing about Bordetella pertussis

The genome of Bordetella pertussis is highly repetitive, containing almost three hundred copies of a 1,053 bp insertion sequence, IS 481.The length of IS 481 means that short-read sequencing, which produces reads of up to only 300 bp, results in genome assemblies comprising many contigs; usually as many contigs as there are copies of IS 481. Yet, the ability of IS 481 to move around the genome, potentially introducing insertions, deletions, rearrangements or duplications, means that the assembly of fully closed B. pertussis genomes may be highly informative. In late 2018 we showed that, using barcoded nanopore sequencing, we can resolve the genomes of several B. pertussis strains per flow cell into closed contigs. However, even when assembled with reads longer than 5,000 bp, the genomes of some strains remained unclosed. We deduced that these unresolvable strains contain ultra-long duplications which, in the same way IS 481 confounds short-read sequencing, are too long to resolve with standard long-read nanopore sequencing.
Here, I highlight recent developments in this work, such as the surprising results of ultra-long nanopore sequencing of these duplication-containing strains: we produced reads in excess of 200,000 bp but were still unable to resolve the genomes, leading to the revelation of highly variable mixed populations. In addition, I will discuss our ongoing work to characterise the most common ultra-long duplication, including quantification of the costs associated with a 5% longer genome, investigation into gene expression differences in the duplicated genes, and attempts to identify phenotypic changes resulting from the duplication. We hope that this exploration of the B. pertussis genome, supported by long-read sequencing, may shed light on previously hidden genotypic and phenotypic variations in this traditionally monomorphic organism.