Optimizing source tracking of Listeria monocytogenes with quasimetagenomics and integrated long and short read sequencing

Abstract Background: Rapid pathogen detection is essential for an effective public health response. The Illumina MiSeq short read sequencer has been the workhorse of many whole genome sequencing source tracking programs. However, short reads cannot span many bacterial genomic repeats, resulting in highly fragmented assemblies. Long read sequencing can span the repeats resulting in the complete assembly of bacterial genomes. With the advancing throughput and accuracy of inexpensive long read sequencing data, it is important to evaluate this resource for potential inclusion in rapid response pathogen identification protocols.Results: Here we sought to maximize the utility of long read (GridIon, Oxford Nanopore) and short read (MiSeq, Illumina) sequencing data for the identification of Listeria monocytogenes from naturally contaminated ice cream. Aliquots from the temporal enrichment (quasimetagenomes) of L. monocytogenes were sequenced and assembled with 10 different assembly approaches. Long read assembly tools generated genome-length contigs and a circularized 71 kbp putative L. monocytogenes plasmid; however, the high sequencing error rate prevented strain typing at even 150X depth of coverage. Short read assemblies provided accurate core gene strain typing after 28 hours of enrichment but were too fragmented for typing the full gene set and reconstructing a circularized genome and plasmid. Hybrid approaches demonstrated the most promising results but were biased by assembly strategy. Short read assemblies scaffolded with long reads were able to accurately strain type L. monocytogenes after just 24 hours of enrichment, but were still too fragmented for sequence typing the full gene set. Long read assemblies polished with short reads recovered genome-length contigs, the full complement of genes, and a circularized plasmid after 24 hours enrichment; however they could not consistently achieve the core gene strain typing accuracy of short read or short read hybrid assembly approaches. Conclusion: These analyses demonstrate that strategic integration and optimization of microbiological (quasimetagenomic), molecular (integrated long and short reads), and bioinformatic approaches (diverse assembly strategies) can greatly improve our ability to quickly and accurately identify pathogens.