All laboratory work with the B. anthracis isolate and nanopore sequencing was performed inside a class II type A2 biological safety cabinet located in a US Federal Select Agent Program registered Biosafety Level 3 laboratory. We performed rapid nanopore sequencing as described by Gargis et al. (13), including silica membrane–based genomic DNA (gDNA) extraction, except that a bead-beating step was added to speed cell lysis. We extracted Ba0914 gDNA from colonies of an overnight agar culture in 75 min. Within the next hour, fluorometer and microvolume spectrophotometer measurements confirmed that the gDNA extraction was suitable for nanopore sequencing. We prepared a nanopore DNA sequencing library (Rapid Barcoding Sequencing Kit SQK-RBK004; Oxford Nanopore Technologies) and sequencing began <45 min later (MinKNOW, version 18.12.6; Oxford Nanopore Technologies).

Within ≈10 min of sequencing, nanopore data were ready for blastn analysis (https://blast.ncbi.nlm.nih.gov), which identified a 13.5-kb read with >91.5% sequence homology with B. anthracis. Approximately 120,000 live basecalled reads (average length 4,089 nt, average quality score/read 14.8) were generated in <5 hours of nanopore sequencing. We performed de novo genome assembly (Flye version 2.5; https://github.com) by using the first 120,000 nanopore reads and error-corrected with Medaka version 0.6.1 (https://pypi.org).

Figure 1

Figure 1. Time required to detect antimicrobial resistance markers in Bacillus anthracis strain Ba0914 by using WGS and summary of assembly results. A) Comparison of time to complete rapid nanopore (MinION) and short-read...

Figure 2

Figure 2. Circular maps of the whole-genome–sequenced Bacillus anthracis Ba0914 chromosome and 2 plasmids, pXO1 and pXO2, assembled by using rapid nanopore sequencing and short-read sequencing. (Maps are not to scale.)

The Ba0914 genome assembly contained single contigs for the chromosome and each plasmid, pXO1 and pXO2 (Figures 1, 2) with >54X average depth of coverage and shared >99.75% identity to the Ames Ancestor strain (GenBank accession no. AE017334) (Table). Thousands of indels and nearly 700 mismatches were detected (Table). Conventional BMD testing also began on August 2, 2019, according to guidelines of the Clinical and Laboratory Standards Institute (14), and susceptibility results were ready the following day.

We also performed Illumina-based SRS. We extracted DNA as described by Gargis et al. (7) and prepared the sequencing library (Nextera DNA Flex Library Kit; Illumina) for paired-end 2 × 150–bp sequencing by using the iSeq 100 (Illumina). We performed read filtering and assembly as described (15). The SRS-based assembly contained 32 more contigs than nanopore but with higher depths of coverage for the chromosome and plasmids (Table). Alignment to the Ames Ancestor reference strain yielded >99.9% genome identity, with fewer indels and mismatches. SRS-based approaches result in lower error rates and can more reliably detect single-nucleotide polymorphisms in B. anthracis, especially in homopolymeric regions, but a same-day laboratory workflow is currently not feasible (Figure 1, panel A). An alignment plot showed gaps in coverage of the chromosome (0.1%) and pXO1 (1.4%) caused by incomplete SRS-based assemblies (Figure 2). Alignment of the nanopore-based assembly to the SRS-based assembly resulted in >99.8% identity but with thousands of indels (Table). All sequencing data were submitted to GenBank under accession no. SAMN12588378.

Only 45 min of bioinformatics analysis (Pima version 01, https://github.com) using 120,000 basecalled nanopore reads was sufficient to assemble and confirm the absence of known AMR genes/markers associated with resistance to quinolones and tetracyclines (Figure 1, panel B) (13). We detected no mutations in gyrA, gyrB, parC, or parE genes, identified no AMR genes contained in the Resfinder database (https://cge.cbs.dtu.dk), and found no unexpected plasmids. SRS and bioinformatics analysis yielded analogous details about the AMR markers in Ba0914 (Figure 1, panel B). Only the SRS-based assembly, and not the nanopore assembly, was reliable for sequencing rsiP. Strain Ba0914 lacked mutations in the homopolymeric rsiP region that can confer penicillin resistance. Sequencing of regions containing repetitive nucleotide bases is a known limitation of nanopore technology and, consequently, detection of the rsiP mutation was excluded from AMR bioinformatics analysis.

Although genetic analysis is useful for detection of known AMR genes/markers in the B. anthracis genome, phenotypic susceptibility testing by BMD remains essential to detect functional resistance (7,13). By using the conventional BMD method, we found that strain Ba0914 was susceptible to ciprofloxacin, levofloxacin, tetracycline, doxycycline, penicillin, and amoxicillin.

Real-time sequencing of the biothreat pathogen B. anthracis in a high-containment laboratory demonstrated the speed and usefulness of a rapid, portable nanopore sequencer during an emergency. Long-read sequencing could detect B. anthracis–specific DNA sequence from the culture isolate after only 3.5 hours. Although the nanopore-based assembly was error-prone when compared with the SRS-based assembly, as few as 8.5 hours would be required to find evidence of known AMR genes/markers or engineering, including gene insertions and extrachromosomal plasmids from B. anthracis. Although conventional AST remains essential for characterizing functional antimicrobial resistance in B. anthracis, nanopore sequencing provided same-day, on-site genomic characterization useful for an anthrax emergency response.