Tonight, I watched the second half of a new Knowledge Exchange session focusing on “Sequencing and analysis of nanopore-only microbial isolates with the NO-MISS workflow.” Different extraction methods produced varying yields. Bead-beating and enzymatic lysis extractions affect read length and throughput, while fungal samples produce lower yields. Some potential issues include incomplete lysis and contaminants. Additional clean-up of samples may be necessary. Twenty-four or ninety-six samples can be prepared with the Rapid Barcoding Kit (RBK) suggested for the NO-MISS workflows. The consideration for batching that was mentioned was the size of the genome. Fungal genomes can be several times larger than bacterial genomes. Running fewer samples can be done in the case of the need for a fast turnaround. The ONT team noted you could load a sample multiple times and wash the flow cell for reuse. A bioinformatics software developer with ONT spoke about improvements in base calling. Updated base calling models now allow for nanopore-only bacterial genome. Key human pathogens can be assembled with high consensus accuracy. To run the EPI2ME wf-bacterial-genomes workflow, a minimum of 32 Gb of RAM and 8 CPUs is needed. I didn’t realize this! For NO-MISS, a minimum of 50X coverage is recommended. The workflow requires FASTQ or BAM files. The isolate mode includes additional features beyond annotation and needs to be activated. Sequencing data is filtered to remove reads <1 kb. I didn’t realize this! Assembly is obtained using Flye, and reads are aligned to the draft genome, which is then polished using Medaka. The initial step of flye is memory intensive. Users can adjust the flye_genome_size and flye_asm_coverage parameters. Prokka is used for annotation, and it uses several tools to annotate the draft: Prodigal, RNAmmer, ARAGORN, SignalP, and Infernal. ONT recommends using the default. MLST is an option within the isolates sub-workflow. Variants in 6-7 housekeeping genes can be analyzed to compare isolates. The MLST tool identifies the appropriate MLST scheme to use. AMR prediction is performed with ResFinder. The presence or absence of ~2,700 genes in the database is checked. Acquired resistance gene detection thresholds can be adjusted. Users can adjust the stringency of matches. The last sub-workflow is serotyping for Salmonella. The output of the wf-bacterial-genomes workflow is an HTML report with interactive graphs. The average depth of each contig is depicted in a graph. Annotation features are saved in two file types: gff and gbk formats, which have the same information. ResFinder saves resistance gene information in a file. This webinar helped me understand the details of the wf-bacterial-genomes workflow!
