Genomic Surveillance of Multi-Drug Resistant Organisms with Long-Read Sequencing

Today was the first day of our IPERT Summer Workshops! I also started watching London Calling 2024 videos available on the YouTube channel. I began with the session titled “Genomic surveillance of multidrug-resistant organisms based on long-read sequencing.” Fabian Landman works at the Centre for Infectious Disease Control in The Netherlands and the Dutch National Institute of Public Health and the Environment. Their goal is to learn about the spread of multi-drug resistant organisms among persons in and between healthcare settings and replace traditional methods with long-read sequencing. Landman and team obtained isolates from May through September 2023 from clinical labs in the countries. They obtained 356 isolates. DNA was extracted using the Promega Maxwell system. The samples were prepared with the Rapid Barcoding Kit v14 (24) and R10.4.1 flow cells. Base calling was performed through Dorado duplex mode for two different modes. They also compared to Illumina output. They binned isolates based on coverage. From >30x coverage, the same number of wgMLST alleles were identified in the Nanopore assemblies compared to the Illumina assemblies. At >40x coverage, “the median number of different wgMLST alleles compared to the Illumina assemblies was relatively stable, between 4 to 7 wgMLST alleles.” The team then compared base calling models. The Rerio base calling model performed the best. They then compared Canu, Fly, Unicycler, and Miniasm. MLST patterns were compared and remained relatively unchanged. AMR gene detection was evaluated and Canu and Flye performed the best for plasmid replicon calling and AMR gene calling specificity. With long-read data, additional genes were identified. Landman noted that the chromosomes were assembled into one contig in almost all cases. Landman also noted that to obtain enough coverage for 40X, multiplexing had to be adjusted when using the MinION and GridION. The P2 allowed multiplexing with the Rapid Barcoding 96 kit, obtaining >130X coverage! They noted this was much more cost-effective. I wonder if they can wash and reuse the PromethION flow cells? Interestingly, Landman concluded that except for Pseudomonas, outbreak analysis with long-read sequencing was possible.

Which long-read assembler and base calling model(s) works best for genomic surveillance analyses? AI-generated image.