Do we still need Illumina sequencing data?: Evaluating Oxford Nanopore Technologies R10.4.1 flow cells and v14 library prep kits for Gram negative bacteria whole genome assemblies

Abstract

The best whole genome assemblies are currently built from a combination of highly accurate short-read sequencing data and long-read sequencing data that can bridge repetitive and problematic regions. Oxford Nanopore Technologies (ONT) produce long-read sequencing platforms and they are continually improving their technology to obtain higher-quality read data that is approaching the quality obtained from short-read platforms such as Illumina. As these innovations continue, we were interested in evaluating how much ONT read coverage produced by the Rapid Barcoding Kit v14 (SQK-RBK114) is necessary to generate high-quality hybrid and long-read-only genome assemblies for a panel of carbapenemase-producingEnterobacteralesbacterial isolates. We found that 30X long-read coverage is sufficient if Illumina data is available, and that 100X long-read coverage is recommended for long-read-only assemblies. We found that Illumina polishing is still improving SNVs and INDELs in long-read-only assemblies. We also examined if antimicrobial resistance genes could be accurately identified in long-read-only data, and found that Flye assemblies regardless of ONT coverage detected > 94 % of resistance genes at 100% identity and length. Overall, the Rapid Barcoding Kit v14 and long-read-only assemblies can be an optimal sequencing strategy depending on the specific use case and resources available.

Hybracter: enabling scalable, automated, complete and accurate bacterial genome assemblies

“…This study also shows that automated perfect hybrid genome assemblies are already possible with Hybracter. This study and others [9,54] also confirm that a long-read first hybrid approach remains preferable to long-read only assembly with Nanopore reads, as short-reads continue to provide accuracy improvements in polishing steps. However, it is foreseeable that short-reads will soon provide little or no accuracy improvements and will not be needed to polish long-read only assemblies to perfection.…”

“…Overall, Hybracter hybrid produced the most accurate chromosome assemblies. For 12 isolates, Hybracter assembled a perfect chromosome (Lerminiaux et al [9] isolates A, B, C, D, G, H, I, J, L, Staphylococcus aureus JKD6159 with R10 chemistry and L. monocytogenes ATCC BAA-679 with simplex and duplex super-accuracy model basecalled reads).…”

“…Medaka long-read polishing was used for all samples except the five ATCC super-accuracy model basecalled duplex read samples, where '--no_medaka' was used. (3) Twelve diverse carbapenemase-producing Gram-negative bacteria from Lerminiaux et al [9]. (4) Staphylococcus aureus JKD6159 sequenced with both R9 and R10 chemistry long-read sets from Wick et al [46].…”

“…As long-read sequencing has improved in accuracy and availability, with the latest Oxford Nanopore Technologies reads recently reaching Q20 (99 %+) median accuracy, a long-read first assembly approach supplemented by short-read polishing has recently been favoured for recovering accurate complete genomes. Long-read first approaches provide greater accuracy and contiguity than short-read first approaches in difficult regions [6][7][8][9][10][11]. The current gold standard manual assembly tool Trycycler even allows for the potential recovery of perfect genome assemblies [7].…”

A novel barcoded nanopore sequencing workflow of high-quality, full-length bacterial 16S amplicons for taxonomic annotation of bacterial isolates and complex microbial communities

“…Depending on the research question at hand, multiplexing capability, sequencing quality, and read depth can be tuned specifically via the given base calling and barcoding options (S1BC, S2BC, D1BC and D2BC). To each option there are noteworthy consequences: Simplex base-calling will greatly increase read depth, but at the cost of sequencing quality, which appears to be around Q20 (25,26). While we can expect high quality reads from duplex base-calling, related workflows will result in less reads per sample.…”