One of the most important aspects of a new laboratory protocol is its reproducibility (as previously discussed on ASM’s Microbial Sciences blog). A protocol must produce the same results when performed by different scientists at different labs, and this is especially important when considering new diagnostic protocols. As whole-genome sequencing (WGS) becomes increasingly used for diagnostic purposes in clinical microbiology labs, the accuracy of WGS-derived diagnoses is important to verify. A new Journal of Clinical Microbiology report demonstrates the reproducibility of accurate results derived by WGS-based typing.
The trial aimed to test the ability of five different clinical labs in three different countries (Denmark, Germany, and The Netherlands) to correctly identify 20 Staphylococcus aureus strains using the Illumina Miseq sequencer. Although each lab used their own standard DNA extraction protocol, all other aspects of data collection, from library prep to data analysis, were standardized, and results were compared to spa typing using Sanger sequencing.
Sequencing entire bacterial genomes can be approached in two ways. One is to investigate allelic changes between the core genome; this is also called core genome multilocus sequence typing (cgMLST); another is to compare the number of single nucleotide polymorphisms (SNPs) between bacterial genomes. The labs sequenced up to 1821 cgMLST alleles for each S. aureus isolate, finding a total of 3 of the 183,927 total cgMLST allele calls were incorrect, or approximately 0.3 incorrect alleles per sample.
Minimum spanning tree comparing sequencing results of 20 different isolates from 5 different clinical labs. Each lab is represented by a different color. Source.
Based on the allelic sequences acquired at each location, the researchers generated a phylogenetic tree showing both relationship of the strains and relationship of the sequencing results (see figure). Almost every allele showed sequence identity between clinical lab locations, and the few differences were still very closely related. In the figure, the numbers on the lines represent the number of alleles differing between locations; while some variation occurred, these variations didn’t affect the overall grouping patterns. This low inaccuracy rate therefore has no effect on analytical outcome and “ensures high intra- and interlaboratory reproducibility,” according to the authors.
The scientists are clear that these results are limited to a single machine and software analysis platform. To broaden their conclusions, different sequencing machinery and analyses would have to be included in future trials. However, this multicenter study sets the stage toward standard use of sequencing methods in clinical microbiology diagnoses.
Photo credit: Illumina MiSeq Image