Fig 1.
Peptidome-based genome comparison pipeline.
Table 1.
The Bacillus strains used in this study. Genome and protein data were retrieved from the BioProject collection of the NCBI in July, 2015.
Fig 2.
Dendrogram showing the hierarchical clustering of all pairwise weighted Robinson-Foulds distances using the Ward method [57].
Four main clusters are found, each one representing one of the four peptidome subsets. The proteome-based GBDP tree that was used as a baseline reference has the smallest distance to the [cytoPI60-more]-based trees and is thus closely positioned to that subgroup. The phylogenetic reconstruction methods FastME, ML, MP and MB are given in parentheses.
Fig 3.
Whole proteome-based phylogenomic tree, including species and subspecies cluster information.
This analysis was based on the GBDP algorithm and rooted with Bacillus subtilis. Numbers above branches are greedy-with-trimming pseudo-bootstrap support values from 100 replicates [13] and only support above 60% is shown. The leaf labels refer to the current NCBI nomenclature, whereas the BCG groups represent the recently revised names [12].
Table 2.
Summary data on the different phylogenetic trees.
Specifically, the average support obtained by the different methods and the size of the character matrices used.
Fig 4.
Bayesian tree based on the [Extracellular] dataset.
Peptides were obtained from extracellular proteins. Bayesian analysis was performed in two independent runs using four Markov chains and 1,000,000 generations [58]. A majority-rule consensus tree (50%) was obtained after discarding the initial 25% of the trees and only support above 60% is shown. The leaf labels refer to the current NCBI nomenclature, whereas the BCG groups represent the recently revised names [12].
Fig 5.
Bayesian tree based on [Cyto28-more] dataset.
It contains peptides with a length higher or equal than 28 amino acids obtained from cytoplasmic. Bayesian analysis was performed in two independent runs using four Markov chains and 1,000,000 generations [58]. A majority-rule consensus tree (50%) was obtained after discarding the initial 25% of the trees and only support above 60% is shown. The leaf labels refer to the current NCBI nomenclature, whereas the BCG groups represent the recently revised names [12].
Fig 6.
Bayesian tree based on [Cyto_PI_51–60] dataset.
It contains peptides with 51–60 amino acids obtained from cytoplasmic proteins with an isoelectric point between 4.5 and 5.5. Bayesian analysis was performed in two independent runs using four Markov chains and 1,000,000 generations [58]. A majority-rule consensus tree (50%) was obtained after discarding the initial 25% of the trees and only support above 60% is shown. The leaf labels refer to the current NCBI nomenclature, whereas the BCG groups represents the recently revised names [12].
Fig 7.
Bayesian tree based on [Cyto_PI_60-more] dataset.
It contains peptides with a length higher than 60 amino acids obtained from cytoplasmic proteins with an isoelectric point between 4.5 and 5.5. Bayesian analysis was performed in two independent runs using four Markov chains and 1,000,000 generations [[58]. A majority-rule consensus tree (50%) was obtained after discarding the initial 25% of the trees and only support above 60% is shown. The leaf labels refer to the current NCBI nomenclature, whereas the BCG groups represent the recently revised names [12].
Table 3.
Running times for classification and protein digestion tasks in Bacillus cereus benchmarking.
Running time is displayed using hours:minutes:seconds format.
Table 4.
Summary data on the NEXUS generation process for Bacillus cereus benchmarking.
Running time is displayed using hours:minutes:seconds format.