bioRxivBioinformaticsPreprint
A capsule polysaccharide synthesis locus database for the Klebsiella oxytoca Species Complex
Melinda M. AshcroftiD, Naoise McGarryiD, Thomas D. StantoniD, Lesley HoylesiD, Kathryn E. HoltiD, Kelly L. WyresiD
Abstract
The Klebsiella oxytoca Species Complex (SC) represents an emerging healthcare-associated group of opportunistic pathogens. The capsular polysaccharide is a virulence determinant and target for novel vaccines, monoclonal antibodies and phage therapy. In the absence of broadly accessible phenotyping techniques, prediction of capsule types from whole-genome sequence data is critical for understanding capsule diversity and epidemiology, and to prioritise capsule types as targets for novel anti-K. oxytoca SC interventions. Here we present the first comprehensive capsule synthesis locus (K locus) database targeted for the K. oxytoca SC, comprising 88 distinct loci defined by gene content and which is compatible with the rapid genome typing tool Kaptive. The database provides high coverage of publicly available K. oxytoca SC genomes (97.6% of 2,244 genomes, dereplicated from a total of 4,055), and the typing rate is significantly higher than that achieved with the pre-existing Klebsiella K locus database (97.6% vs 50.3%, p <0.0001), which primarily targets the Klebsiella pneumoniae SC. We demonstrate the utility of the novel K. oxytoca SC database by application to three diverse clinical K. oxytoca SC isolate collections (n=61 to 102 genomes each), suggesting a high diversity of K types. The novel K. oxytoca SC K locus database (github.com/klebgenomics/KoSC-surface-antigen-loci) will provide a key resource to support larger systematic studies and ongoing genomics surveillance efforts for the K. oxytoca SC. IMPACT STATEMENTMembers of the Klebsiella oxytoca Species Complex (SC) are an emerging cause of infections in humans and are frequently associated with antimicrobial resistance. Klebsiella species produce two key surface antigen sugars (capsular polysaccharide and lipopolysaccharide) that are immunogenic and are targets for novel control strategies such as vaccines and phage therapy. Phenotypic typing of these surface antigen sugars (serotyping) is costly and laborious, with genotyping (predicting the serotype from whole-genome sequence data) a useful alternative. Here, we present a curated capsule (K) locus reference database for the K. oxytoca SC, which represents a useful tool to assist in the epidemiological surveillance of this emerging pathogen. DATA SUMMARYAll Klebsiella oxytoca Species Complex genomes used in this work were publicly available, with accession details listed in Supplementary Tables 1, 2 and 3. The K. oxytoca Species Complex K locus reference database is available under GNU public license at github.com/klebgenomics/KoSC-surface-antigen-loci.
- organism: bacteria (other)
- assay: whole-genome sequencing
- research focus: database / resource
- approach: statistical model
- approach: benchmark study
In-depth summary
Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (17 Jul 2026) · CC BY 4.0
A curated Kaptive-compatible database of 88 capsule (K) loci types 97.6% of dereplicated Klebsiella oxytoca species complex genomes, versus 50.3% typed with the existing K. pneumoniae database.
Other teams applying the K. pneumoniae species complex K locus database to K. oxytoca species complex genomes reported typeability of 0-33%, and Kaptive can mistype genomes whose loci are missing from its reference database. This work builds a species-complex-specific locus reference from public genomes and tests how well it covers them.
Key results
ShownThe database holds 88 distinct K loci defined at an 82.5% translated amino acid identity threshold. All loci carry galF, cpsACP, wza, wzb, wzc, gnd and ugd, while wzi, wzx and wzy were detected in only 87, 86 and 82 loci.
Locus length 21,171 bp (KL37) to 30,521 bp (KL40), median 25,901 bp; 17 to 25 ORFs. wzi in 87 loci (absent in KL81), wzx in 86, wzy in 82. Initiating glycosyltransferase: wbaP in 77 loci, wcaJ in 11. · K. oxytoca Species Complex K locus database…
ShownThe new database types far more K. oxytoca species complex genomes than the K. pneumoniae species complex database, and reports more distinct loci.
3,981 of 4,055 genomes (98.1%) typeable. In the dereplicated set of 2,244 genomes, 97.6% typed vs 50.3% with the K. pneumoniae database (Fisher's exact p < 0.0001, OR 41.6, 95% CI 31.14-56.59); 88 vs 22 loci reported. · K. oxytoca SC K locus database enables…
SuggestedRarefaction curves suggest the database covers most loci in the available genomes. Coverage is clearest for K. oxytoca, and the curves for other species are still rising.
K. oxytoca (n=689) plateaued at 36 loci. K. michiganensis (n=1,088) and K. grimontii (n=348) were trending toward plateau at >40 and >50 loci. K. pasteurii (n=81) had not plateaued. · K. oxytoca SC K locus database enables…; Figure legends
ShownMany K. oxytoca species complex loci are orthologous to K. pneumoniae species complex loci, so some capsule structures can be predicted by orthology. Locus frequencies are strongly skewed, with most loci rare.
10 loci completely orthologous and 7 distant orthologs. 31% of typeable dereplicated genomes (n=674) carry a direct or likely functionally equivalent ortholog, and 16% (n=346) a distant one. 54 loci (61%) occur in <1% of typeable genomes; 14 (16%) in >2%. · K. oxytoca Species Complex K locus database…; K loci frequency and species distribution
SuggestedIn three clinical isolate collections (Australia, China, Italy) every genome was typed, showing high locus diversity and a partly shared set of common loci. The authors call this illustrative only.
All 255 genomes typed; 23-38 distinct loci per study and 54 of 88 loci seen overall. KL74, KL34 and KL83 together accounted for 28.2% of isolates. KL83 was the commonest locus in Italy (n=12 of 61). · Use case: seroepidemiology of clinical K…; DISCUSSION
Methods, data and reproducibility
K loci were extracted from public K. oxytoca species complex assemblies with Kaptive, clustered, annotated and compared by gene content at 82.5% amino acid identity, then curated into a Kaptive reference database. Typing performance was tested on 4,055 genomes, with a dereplicated subset and a comparison against the K. pneumoniae database.
- Preliminary dataset of public K. oxytoca species complex assemblies (Pathogenwatch, NCBI, published studies) after quality control, used to define the initial loci · Klebsiella oxytoca species complex · n = 3,288 assemblies
- Final dataset from NCBI, Pathogenwatch and AllTheBacteria (as of 5 April 2025), used for database testing · Klebsiella oxytoca species complex (plus K. indica) · n = 4,055 assemblies
- Dereplicated subset of the final dataset used to report typing performance and rarefaction · Klebsiella oxytoca species complex · n = 2,244 genomes
- Published clinical isolate collection, Australia · Klebsiella oxytoca species complex · n = 92
- Published clinical isolate collection, China · Klebsiella oxytoca species complex · n = 102
- Published clinical isolate collection, Italy · Klebsiella oxytoca species complex · n = 61
- Methods & tools
- Kaptive, Kleborate, MeShClust, Prokka, MMseqs2, HMMER, Unicycler, Flye, MetaPhlAn, Mash, vegan, clinker
- Data & accessions
- https://github.com/klebgenomics/KoSC-surface-antigen-loci
- Software & environment
- Tool versions are given: Kaptive v2.0.3 (locus extraction) and v3.1.0 (testing), Kleborate v3.2.4, Mash v2.3, vegan v2.7.2, clinker v0.0.32, R v4.5.1.
- Assay conditions & compute
- Monash eResearch computing resources (M3) are acknowledged; no runtimes or memory are reported.
- Reproducibility gaps
- Genome accessions are in Supplementary Tables 1-3, not in the main text; no per-step runtimes are given.
Limitations (5)
- The dereplicated dataset is a convenience sample of public genomes, biased toward human isolates (70% where source was known) and high-resource settings, so locus frequencies may not reflect true population frequencies. · stated by the authors
- The three clinical collections are small and used different sampling strategies, so the seroepidemiology results are illustrative; sample sizes (n≤8 genomes per ST per dataset) were too small for statistical comparison. · stated by the authors
- Species restriction of KL34 and KL69 to K. oxytoca sensu stricto may be an artefact of the convenience sample rather than a genuine restriction. · stated by the authors
- The database was built from the preliminary dataset, and 8 more loci were added after untypeable genomes in the 4,055-genome test set were examined. The test set also includes the 255 clinical genomes. No independent genome set is reported, so the typing rates are partly in-sample. · noted in this summary
- Capsule structure predictions rest on orthology to K. pneumoniae loci with known structures. Most loci have no described polysaccharide structure or serological validation. · noted in this summary
Read this if you type Klebsiella oxytoca species complex genomes with Kaptive, or need a curated K locus reference for surveillance of this group.
- Citations
- 1
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Kaptive 2.0: updated capsule and LPS locus typing for the Klebsiella pneumoniae species complex
Margaret M. C. Lam, Ryan R. Wick, … Kelly L. Wyres·bioRxiv·Nov 5, 2021
- Identification of Klebsiella capsule synthesis loci from whole genome data
Kelly L. Wyres, Ryan R. Wick, … Kathryn E. Holt·Microbial GenomicsPeer-reviewed·Dec 1, 2016
- Fast and Accurate in silico Antigen Typing with Kaptive 3
Thomas D. Stanton, Marit Andrea Klokkhammer Hetland, … Kelly L. Wyres·Microbial GenomicsPeer-reviewed·Jun 24, 2025
- Genome-resolved surveillance of African Klebsiella oxytoca species complex genomes reveals resistome-mobilome and biosynthetic gene cluster diversity
Samweli Y. Bahati, Abdalah Makaranga, … Reuben S. Maghembe·bioRxiv·Aug 26, 2026
- Kaptive Web: user-friendly capsule and lipopolysaccharide serotype prediction for Klebsiella genomes
R. R. Wick, E. Heinz, … K. L. Wyres·bioRxiv·Feb 5, 2018
- Improved molecular characterization of the Klebsiella oxytoca complex reveals the prevalence of the kleboxymycin biosynthetic gene cluster
P. Shibu, F. McCuaig, … L. Hoyles·bioRxiv·Jul 24, 2020