Influence of empirically derived filtering parameters, amplicon sequence variant, and operational taxonomic unit...

Microbes play an important role in human and animal health, as well as animal productivity. The host microbial interactions within ruminants play a critical role in animal health and productivity and provide up to 70% of the animal's energy needs in the form of fermentation products. As such, many studies have investigated microbial community composition to understand the microbial community changes and factors that affect microbial colonization and persistence. Advances in next-generation sequencing technologies and the low cost of sequencing have led many studies to use 16S rDNA-based analysis tools for interrogation of microbiomes at a much finer scale than traditional culturing. However, methods that rely on single base pair differences for bacterial taxa clustering may inflate or underestimate diversity, leading to inaccurate identification of bacterial diversity. Therefore, in this study, we sequenced mock communities of known membership and abundance to establish filtration parameters to reduce the inflation of microbial diversity due to PCR and sequencing errors. Additionally, we evaluated the effect of the resulting filtering parameters proposed using established bioinformatic pipelines on a study consisting of Holstein and Jersey cattle to identify breed and treatment effects on the bacterial community composition and the impact of filtering on global microbial community structure analysis and results. Filtration resulted in a sharp reduction in bacterial taxa identified, yet retain most sequencing data (retaining >79% of sequencing reads) when analyzed using 3 different microbial analysis pipelines (DADA2, Mothur, USEARCH). After filtration, conclusions from α-diversity and β-diversity tests showed very similar results across all analysis methods. The mock community-based filtering parameters proposed in this study help provide a more realistic estimation of bacterial diversity. Additionally, filtration reduced the variation between microbiome analysis methods and helped to identify microbial community differences that could have been missed due to the large animal-to-animal variation observed in the unfiltered data. As such, we believe the new filtering parameters described in this study will help to obtain diversity estimates that are closer to realistic values, improve the ability to detecting microbial community differences, and help to better understand microbial community changes in 16S rDNA-based studies.