Rapid draft sequencing and real-time nanopore sequencing in a hospital outbreak of Salmonella

Abstract

Background: Foodborne outbreaks of Salmonella remain a pressing public health concern. We recently detected a large outbreak of Salmonella enterica serovar Enteritidis phage type 14b affecting more than 30 patients in our hospital. This outbreak was linked to community, national and European-wide cases. Hospital patients with Salmonella are at high risk, and require a rapid response. We initially investigated this outbreak by whole-genome sequencing using a novel rapid protocol on the Illumina MiSeq; we then integrated these data with whole-genome data from surveillance sequencing, thereby placing the outbreak in a national context. Additionally, we investigated the potential of a newly released sequencing technology, the MinION from Oxford Nanopore Technologies, in the management of a hospital outbreak of Salmonella.

Results: We demonstrate that rapid MiSeq sequencing can reduce the time to answer compared to the standard sequencing protocol with no impact on the results. We show, for the first time, that the MinION can acquire clinically relevant information in real time and within minutes of a DNA library being loaded. MinION sequencing permits confident assignment to species level within 20 min. Using a novel streaming phylogenetic placement method samples can be assigned to a serotype in 40 min and determined to be part of the outbreak in less than 2 h.

Conclusions: Both approaches yielded reliable and actionable clinical information on the Salmonella outbreak in less than half a day. The rapid availability of such information may facilitate more informed epidemiological investigations and influence infection control practices.

Fig. 1

Phylogenetic reconstruction of the outbreak, as sequencing data were generated at three time points during the outbreak (corresponding to Panels a, b and c). The red node demonstrates the most frequently occurring SNP type. Node labels show the number of isolates of this type. Blue nodes are singletons. Edge labels show the number of SNPs between isolates. The turquoise node shows a cluster of two identical isolates from cases associated with a nursery. The header bar shows the date of the sequencing, the Illumina sequencing protocol used, the number of samples in the dataset and the total percentage of the S. Enteritidis reference genome which could be used for SNP calling (core genome). Subtle differences in the number of SNPs between nodes on the tree may be seen, for example in the unrelated nursery isolates (6 SNPs different between Panel a and Panel b). These differences are attributed to differences in core genome size, that is, the number of positions in the reference genome used to generate the results. The precise core genome size figures are shown as a percentage in the panel headings

Fig. 2

Phred-scaled quality scores (−10 log₁₀ P) for Illumina sequencing demonstrating the impact of read length on read quality scores with the three Illumina MiSeq sequencing modalities used in this study. Red scores indicate results from the draft 1 × 75 base sequencing protocol, which shows minimally worse quality drop-off than running V2 (green, 2 × 250 base) or V3 chemistry (blue, 2 × 300) under standard conditions

Fig. 3

Streaming taxonomic assignments from the Oxford Nanopore MinION are shown for an isolate belonging to the outbreak and an isolate not belonging to the outbreak. Assignments to Salmonella and Salmonella enterica are found within 10 min of starting sequencing for the outbreak strain and within 20 min for the non-outbreak strain. Phage-specific sequences are detected and are distinct between the non-outbreak and outbreak strain. The non-outbreak strain harbours Salmonella phage ST64B and Gifsy-2, whereas the outbreak strain harbours Salmonella phage RE2010

Fig. 4

Results of streaming phylogenetic placement from the Oxford Nanopore MinION on a reference tree of representative published Salmonella enterica sequences. Common serovars of Salmonella are highlighted. Both outbreak and non-outbreak strains are unambiguously identified as Salmonella enterica serovar Enteritidis by their position on the phylogenetic tree within 50 min. The line demonstrates the potential range of placements reported by pplacer. The red placements indicate the positions of the outbreak isolate and the blue placements indicate the positions of the non-outbreak isolate

Fig. 5

Results of streaming phylogenetic placement from the Oxford Nanopore MinION on a reference tree of Salmonella enterica serovar Enteritidis isolates collected by Public Health England during routine surveillance. The left-most panel demonstrates the confident placing of the outbreak isolate in the outbreak clade within 100 min, and the confident placing of the non-outbreak isolate into a clade containing multiple serotypes of Salmonella within 120 min. The red placements indicate the positions of the outbreak isolate and the blue placements indicate the positions of the non-outbreak isolate. The right-most panel shows a phylogenetic reconstruction of isolates from the outbreak and their source, set in context of a national outbreak of phage type 14b. Uncertainty in the phylogenetic placement technique is demonstrated early on in sequence data collection due to the low accuracy of the variant calls collected. As more data are collected, the number of possible phylogenetic placements reduces and the confidence values increase (not shown)

Fig. 6

A schematic showing time-to-result for the sequencing analysis strategies used in this study, starting from DNA input