Acinetobacter genome assemblies from our collection for which the KL and OCL types had been previously determined via manual or automated sequence inspection ([2, 41 (link)]; and unpublished data) were used to assess the level of typing accuracy that could be achieved through the use of our novel databases with Kaptive. Paired-end Illumina read data (described in [2, 41 (link)] and available under BioProject accession PRJEB2801) were de novo assembled using SPAdes v. 3.13.1 [46 (link)] and optimized with Unicycler v. 0.4.7 [47 (link)]. High-quality genome assemblies (n=719) with a maximum contig number of 300 and minimum assembly length of 3.6 Mbp were included in the analysis (cut-offs determined empirically by manual inspection of the contig number and assembly length distributions, respectively). These assemblies were assessed for oxaAb presence using blastn (>95 % nucleotide sequence similarity and >90 % combined coverage) to confirm the A. baumannii species assignment. Confirmed A. baumannii sequences (n=642) were analysed using both KL and OCL reference databases with command-line Kaptive v. 0.7 [44 (link)] with default parameters.
The same method was used to test databases against 3412 genome sequences available in the NCBI non-redundant and WGS databases as of February 2019. These genome assemblies were bulk downloaded from NCBI as a compressed .tar file for local analysis. Genomes lacking oxaAb were removed prior to typing but quality control (QC) analysis as described above was applied to this data set only after typing was complete.
Free full text: Click here