A new family of bacterial DNA repair proteins annotated by the integration of non-homology, distant homology and structural bioinformatic methods

AbstractDifferent bioinformatics methods illuminate different aspects of protein function, from specific catalytic activities to broad functional categories. Here, a triple-pronged approach to predict function for a domain of unknown function, DUF2086, is applied. Distant homology to characterised enzymes and conservation of key residues suggest an oxygenase function. Modelling indicates that the substrate is most likely a nucleic acid. Finally, genomic context analysis linking DUF2086 to DNA repair, leads to a predicted activity of oxidative demethylation of damaged bases in DNA. The newly assigned activity is sporadically present in phyla not containing near relatives of the similarly active repair protein AlkB