Bacillus subtilis, the model Gram-positive bacterium: 20 years of annotation refinement

Abstract

Genome annotation is, nowadays, performed via automatic pipelines that cannot discriminate between right and wrong annotations. Given their importance in increasing the accuracy of the genome annotations of other organisms, it is critical that the annotations of model organisms reflect the current annotation gold standard. The genome of Bacillus subtilis strain 168 was sequenced twenty years ago. Using a combination of inductive, deductive and abductive reasoning, we present a unique, manually curated annotation, essentially based on experimental data. This reveals how this bacterium lives in a plant niche, while carrying a paleome operating system common to Firmicutes and Tenericutes. Dozens of new genomic objects and an extensive literature survey have been included for the sequence available at the INSDC (AccNum AL009126.3). We also propose an extension to Demerec's nomenclature rules that will help investigators connect to this type of curated annotation via the use of common gene names.

Figure 1

Based on SubtiList, a draft interface for microbial databases built up for tablets at the BGI.

Figure 2

Scenarios for annotation. Annotation combines three approaches: data‐, hypothesis‐ and context‐driven. The first one is based on induction, the second on deduction and the third on abduction, combining functional, phenotypic and sequence data (orange boxes and see text). The outcome of the procedure results in the identification of a gene product, a gene name, participation in metabolic reactions and literature references identified by PubMed identifiers (black boxes). Free text notes are also provided to help understanding the biologically relevant context of each particular gene.

Figure 3

Analysis of protein‐coding genes in 36 complete genomes of B. subtilis.A. The core and pan‐genomes were computed for random samples of increasing size of the 36 genomes. The shaded regions indicate the range of variation of these values.B. The frequency of the presence of each gene family from those that are present in only one strain (peak at 1) to those that are components of the core genome (peak at 36). The identification of the families of core and pan‐genomes followed the methodology of (Touchon et al., 2014).