A New Genomics-Driven Taxonomy of Bacteria and Archaea: Are We There Yet?

Abstract

Taxonomy is often criticized for being too conservative and too slow and having limited relevance because it has not taken into consideration the latest methods and findings. Yet the cumulative work product of its practitioners underpins contemporary microbiology and serves as a principal means of shaping and referencing knowledge. Using methods drawn from the field of exploratory data analysis, this minireview examines the current state of the field as it transitions from a taxonomy based on 16S rRNA gene sequences to one based on whole-genome sequences and tests the validity of some commonly held beliefs.

FIG 1

Principal-component analysis of 16S rRNA sequences from type strains of validly named species and subspecies of Bacteria and Archaea. PCA was used to visualize the distribution pattern of 16S rRNA sequence in a manner similar to that previously described by Garrity and Lilburn (25) with the following modifications. High-quality, curated sequences, matched to the published taxonomic descriptions of type strains, were aligned using the most recent Greengenes alignment, and uncorrected pairwise evolutionary distances were determined in Mothur (version 1.34.1) using the default parameters (Needleman-Wunsch method, kmer searching with 8 mer). The resulting matrix was then subjected to a PCA in R (version 3.3.2, 64 bit, Mac OS). The first three principal components accounted for 86.9% of the total variance. Areas highlighted in red indicate locations of members of the two prokaryotic domains. The two outliers located in the lower right quadrant are Methanocorpusculum sinense (GenBank accession no. AF095268.1) and Halostagnicola bangensis (HF544345.1). Additional PCA plots showing the locations of the validly named orders and classes are available in Fig. 1S posted at https://www.researchgate.net/publication/302545989_JCM00200-16-supplement. A complete set of figures is available for viewing at http://dx.doi.org/10.1601/tx.1.

FIG 2

Distribution of sequence similarity among taxon members. The distribution of 16S sequence similarity was determined from the output of the SOSCC analysis of the complete matrix (n = 11,814) for each taxon for the rank of genus to domain, excluding singletons. Pairwise similarities were extracted from the lower triangle of the matrix (excluding self-identity scores on the diagonal). The five number summaries (mean, median, maximum, and first and third quartiles) (11) were computed for each taxon, and boxplots summarizing the distribution of scores at each rank were drawn. Output of the SOSCC is provided in Table S2 posted at https://www.researchgate.net/publication/302545989_JCM00200-16-supplement.