Metagenomics or Metataxonomics: Best Practice Methods to Uncover the Sinus Microbiome

Abstract Background Dysbiosis of the human microbiome has been linked to many chronic diseases including chronic rhinosinusitis (CRS). Advances in next generation sequencing have improved our ability to identify difficult-to-culture bacteria, many of which populate the sinuses. However, methods of microbiome analysis have not been benchmarked in the sinuses, and sequencing workflows have been borrowed from more extensively studied environments such as the gut microbiome. Currently, the gold-standard method used by most researchers to analyse the sinonasal microbiome is 16s rRNA sequencing. However, despite following best practice, significant discrepancies in microbiome results are reported in the literature. Although differences in demographics, sample size, sampling techniques, library preparation and bioinformatic analysis may in part explain this, the reliability of 16s rRNA sequencing itself for sinus microbiome analysis is yet to be examined. Methods In this study we create the first sinus-relevant mock-community and use this as a positive control to benchmark genomic methods of analysis for sinus microbiome study. The mock community was assembled with equal proportions of 9 strains of bacteria common to the sinuses. Five different library preparation/sequencing methods were employed to generate 29 unique samples. Taxonomic profiles were generated with emu for the long read (LR) 16S datasets, dada2/SILVA for the short read (SR) 16s datasets and sourmash for the metagenomic datasets. Results We believe the results of this study mark a turning point in sinus microbiome research. Our work shows that 16s sequencing, the current gold standard method of analysing sinus microbiomes, produces unrecognisable results when compared to the ground truth; and that this distortion of results is both PCR-primer and species specific. 16S rRNA PCR amplification introduces excessive bias and thus, subsequent taxonomic profiling is misrepresentative of the input microbiome. This was consistent for SR and LR 16s rRNA sequencing. By contrast, SR and LR shotgun metagenomic sequencing was able to, repeatedly and accurately, recapitulate the taxonomic profile of the input mock community. When we applied these methods to a patient sample, we saw a dramatic difference in the taxonomic profile of the microbiome, with shotgun sequencing revealing the dominance of Corynebacterium spp.. Conclusions In order to reach meaningful conclusions that impact clinical practice and improve patient outcomes, we need reliable and robust methods. In this study we found that methods validated in the context of the gut microbiome performed poorly when applied to the sinus microbiome, highlighting the need for appropriate, ecology-specific benchmarking. This work finds that shotgun metagenomic sequencing is the most accurate method with which to analyse the sinus microbiome. Future studies in the sinus microbiome should use shotgun sequencing where relative abundance as well as taxonomy is relevant, other methods are not fit for this purpose. LR 16s rRNA sequencing with the KAPA primer is able to detect relevant bacterial species but does not accurately represent relative abundance.