Abstract
<jats:p>Acinetobacter baumannii is one of the most critical bacterial pathogens requiring novel approaches for infection control and treatment to curb the spread of highly resistant isolates. Epidemiological surveillance, and the design and application of many non-antibiotic interventions, require rapid and accurate prediction of diverse capsular polysaccharide (CPS) types from whole genome sequence data. The internationally adopted CPS typing system relies on the in-silico detection of CPS biosynthesis genes both in and outside the K locus (KL), utilising a reference sequence database that is compatible with the bioinformatics tool, Kaptive. In this study, 168 novel loci were added to the database following an extensive survey of publicly available A. baumannii genomes and non-redundant sequence entries, bringing the total number of reference sequences to 409 KL and 10 extra-locus genes. All novel loci conformed to the characteristic K locus configuration described previously, with conserved core genes for CPS biosynthesis flanking a central and highly variable region that includes type-specific structural genes. Across the 409 KL, there were 1000 protein clusters. Annotations for 309 novel clusters were curated and validated using a variety of sequence- and structure-based approaches. Most proteins (n=781) were encoded by genes found in <4 KL, consistent with extensive structural diversity of CPS types in A. baumannii as reported previously. However, gene distribution analysis predicted shared structural features, with several KL predicting the same CPS type or K unit. Validation of the updated database against 46,185 publicly available genomes revealed that 32 K loci were present in 87.1% of sequenced isolates, with an overrepresentation of genomes with KL2, KL3, KL18, KL9 or KL22.</jats:p>