A reference dataset of bacterial 16S ribosomal gene V4 amplicon sequences obtained from gastro-intestinal samples of farm animal species.. bacteria

Sequencing of bacterial 16S ribosomal gene subunit variants has been well established in literature as a method of surveying the taxonomic structure of bacterial communities in environmental samples, but poor taxonomic specificity of sequence clusters hinders efforts to relate independent sequence datasets. Here, we propose that the use of a defined, reference sequence dataset in which all sequences have previously been determined to maintain at least 3.0 percent autonomy can improve specificity and repeatability of sequence analysis results. Samples of gastro-intestinal material (n = 298) from bovine rumen, bovine rectum, ovine rumen, ovine rectum, and avian cecum were subjected to DNA extraction procedures and the hypervariable V4 region of the bacterial 16S ribosomal gene was amplified by multiplex PCR and sequenced by Illumina MiSeq. Sequences were initially processed by the MOTHUR pipeline and then were refined to select only sequences that maintained at least 3.0 percent dissimilarity from one another on the basis of pairwise alignment. The resulting collection of 16S V4 amplicon sequences (n = 5,489) and supporting data were submitted as a reference sequence dataset.The reference sequence dataset was used to produce a multiple sequence alignment and was subjected to clustering and taxonomic assignment procedures in MOTHUR. The results demonstrate that separation of sequences by 97.0 percent similarity on the basis of pairwise alignment was not synonymous with clustering of sequences on the basis of multiple sequence alignment with either furthest neighbor, nearest neighbor, or average neighbor algorithms. In addition, the results demonstrate that bacterial families such as Lachnospiraceae and Ruminococcaceae, as well as other numerous bacterial taxonomic groups, can be analyzed with greater specificity then when only described by consensus taxonomy. We therefore propose the use and continued development of the reference sequence dataset to augment conventional sequence analyses that are based on sequence clustering.