Assessment of metrics in next-generation sequencing experiments for use in core-genome multilocus sequence type

With the reduction in the cost of next-generation sequencing, whole-genome sequencing (WGS)-based methods such as core-genome multilocus sequence type (cgMLST) have been widely used. However, gene-based methods are required to assemble raw reads to contigs, thus possibly introducing errors into assemblies. Because the robustness of cgMLST depends on the quality of assemblies, the results of WGS should be assessed (from sequencing to assembly). In this study, we investigated the robustness of different read lengths, read depths, and assemblers in recovering genes from reference genomes. Different combinations of read lengths and read depths were simulated from the complete genomes of three common food-borne pathogens: Escherichia coli, Listeria monocytogenes, and Salmonella enterica. We found that the quality of assemblies was mainly affected by read depth, irrespective of the assembler used. In addition, we suggest several cutoff values for future cgMLST experiments. Furthermore, we recommend the combinations of read lengths, read depths, and assemblers that can result in a higher cost/performance ratio for cgMLST.

Automated summary

This study evaluated the robustness of different read lengths, read depths, and assemblers in recovering genes from reference genomes. The quality of assemblies was found to be mainly affected by read depth, regardless of the assembler used. Recommendations for cutoff values and cost-effective combinations for cgMLST were provided.