[en] My aim is to produce possible scenarios for the bacterial evolution based on the bacterial
phylogeny and the bacterial cell-wall. For that, we need a selection of genomes which represent
the bacterial diversity and are not redundant. However, there is an overabundance of bacterial
genomes and most are redundant, so a solution to remove redundant genomes while conserving
the bacterial diversity was needed. Yet, none were available when I began my thesis.
I created a tool to automatically cluster genomes and select the best representative for each
cluster. The clustering is based on whole genome comparison and the selection considers
genome quality, annotation richness, completeness level and absence of contamination. We
called my tool ToRQuEMaDA (Tool for retrieving queried Eubacteria, metadata and dereplicating
assemblies) or TQMD for short. TQMD is optimized to dereplicate at high taxonomic levels
(phylum) but remains competitive while compared to other programs which are optimized to
dereplicate at low taxonomic levels (species).
Based on a selection of 903 genomes, we computed orthologous groups (OGs) from which we
studied the synteny of the division and cell wall (dcw) cluster. Using a smaller selection of
genomes, 85, we produced a phylogenomic tree based on the 117 most conserved (and single
copy) genes in our selection of bacterial genomes. Using this tree, we reconstructed the dcw
cluster using an ancestral gene order reconstruction tool and the last bacterial common ancestor
(LBCA) cell wall using Bayesian Inference. From our results, it appears that the LBCA was a
monoderm already featuring a peptidoglycan layer. We further studied genes involved with the
outer membrane (OM) to validate (or invalidate) our results and did not find decisive clues to reject
them.
Research Center/Unit :
CIP - Centre d'Ingénierie des Protéines - ULiège Biological Sciences from Molecules to Systems - inBioS