An inter-laboratory study to investigate the impact of the bioinformatics component on microbiome analysis

BackgroundDespite the advent of whole genome metagenomics for microbiome analysis, targeted approaches (such as 16S rRNA amplicon sequencing) continue to offer valuable methods for also determining the microbial composition of a sample. In this context a sample can be referring to a clinical sample from a normally sterile site where the requirement is to determine the aetiology of the infection (usually single pathogen identification) or to sampling from more complex samples such as human mucosal organs or the environment where multiple micro-organisms need to be identified. The methodologies are frequently applied to determine both presence of micro-organisms and their quantity. There are a number of technical steps required to perform a bacterial community profiling protocol many of which may have appreciable precision and bias that will impact on the final result. In order for these methods to be applied with the greatest accuracy, comparative studies across different laboratories are warranted. ResultsIn this study we explored the impact of the bioinformatic approaches taken in different laboratories on microbiome assessment using 16S rRNA amplicon sequencing results. Data was generated from two mock microbial community samples which were amplified using primer sets spanning five different variable regions of the 16S rRNA gene. The analysis included three technical repeats of the process to determine the repeatability of their methods. Thirteen laboratories participated in the study where they analysed the same FastQ files using their choice of pipeline. The study captured the methods used and the corresponding abundance of bacteria in each of the samples. Results were compared to the digital PCR assessment of the absolute abundance of each target representing each organism in the mock microbial community samples and also to the shotgun metagenomic sequencing approach estimating community composition. ConclusionsThis study demonstrated that the choice of bioinformatic analysis pipeline for 16S rRNA amplicon sequencing experiments can have a notable impact on the results generated. The study observed differences in terms of presence and abundance of organisms and provides a resource for pipeline development. The observed differences were especially prevalent when using custom databases and applying high threshold cut-off limits. In order to apply sequencing approaches with greater accuracy and to underpin their routine implementation, we need to deepen our understanding of the impact of different analytical steps and to devise solutions for harmonising microbiome analysis results.