Several novel non-antibiotic therapeutics for the critical priority bacterial pathogen, Acinetobacter baumannii, rely on specificity to the cell-surface capsular polysaccharide (CPS). Hence, prediction of CPS type deduced from genes in whole genome sequence data underpins the development and application of these therapies. In this study, we provide a comprehensive update to the A. baumannii K locus reference sequence database for CPS typing (available in Kaptive v. 2.0.1) to include 145 new KL, providing a total of 237 KL reference sequences. The database was also reconfigured for compatibility with the updated Kaptive v. 2.0.0 code that enables prediction of 'K type' from special logic parameters defined by detected combinations of KL and additional genes outside the K locus. Validation of the database against 8994 publicly available A. baumannii genome assemblies from NCBI databases identified the specific KL in 73.45 % of genomes with perfect, very high or high confidence. Poor sequence quality or the presence of insertion sequences were the main reasons for lower confidence levels. Overall, 17 KL were overrepresented in available genomes, with KL2 the most common followed by the related KL3 and KL22. Substantial variation in gene content of the central portion of the K locus, that usually includes genes specific to the CPS type, included 34 distinct groups of genes for synthesis of various complex sugars and >400 genes for forming linkages between sugars or adding non-sugar substituents. A repertoire of 681 gene types were found across the 237 KL, with 88.4 % found in <5 % of KL.
Keywords: Acinetobacter baumannii; K locus; KL; Kaptive; capsular polysaccharide.