henhao
Sign in
MicrobiologybioRxiv
Preprint

Phylogeny informed international clone assignment with PhyloMLST

Matthew NeiliD, Benjamin Andrew EvansiD

Abstract

Multilocus sequence typing (MLST) remains the predominant method for typing bacterial strains. A common method for investigating particularly successful epidemic lineages within a species is to cluster isolates with similar MLST profiles into clonal complexes (CCs). Some CCs, such as international clones (ICs) in A. baumannii, are identified with specific sequence types (STs) and are of particular importance to human health. Although theoretically simple, there is a lack of convenient, user-friendly tools perform this analysis. Here we present PhyloMLST, a tool to cluster bacterial isolates into CCs and map them to ICs using the output from existing MLST tools and user-provided STs. As there is potential for aberrant IC assignment arising from excessively large CCs constructed with spurious links, PhyloMLST provides additional functionality to correct IC assignment with a user-provided phylogenetic tree. Although designed with A. baumannii in mind, PhyloMLST can be applied to any bacteria where construction and investigation of CCs based on MLST is performed. Impact statementMany bacterial pathogens are characterised by successful epidemic lineages that are responsible for a substantial number of infections, may be more virulent, and may carry an abundance of antimicrobial resistance genes. These epidemic lineages are comprised of a number of multilocus sequence typing (MLST) sequence types (STs), clustered into clonal complexes (CCs). To date, identifying which STs belong to which epidemic lineage has been challenging, with no simple analytical tools available. Here, we present PhyloMLST - a phylogenetically-aware method for assigning STs to epidemic lineages. The customisable nature of the tool will enable researchers to straightforwardly characterise any population of bacteria that they are working on using MLST data and user-defined definitions of epidemic lineages. Data summaryThe PhyloMLST source code and example data shown here is available at https://github.com/Mattn286/PhyloMLST.

PDFDOICodeData
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
0
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (23 Jul 2026) · CC BY 4.0

PhyloMLST, a Python command-line tool, groups MLST sequence types into clonal complexes and international clones, optionally using a phylogeny to remove spuriously chained 'straggly' assignments.

Few command-line tools link MLST sequence types into clonal complexes: PubMLST provides them only for some species and from its whole database, and goeBURST in PHYLOViZ needs Java and drops novel profiles, so A. baumannii studies often treat single STs as proxies for international clones. Chaining STs through intermediates also yields 'straggly' complexes, which, per the authors, no prior method corrects with a phylogeny.

Key results

  1. ShownPhyloMLST, a dependency-light Python 3 command-line tool, builds clonal complexes (CCs) at a user-chosen locus-variant level, maps CCs to international clones (ICs) through user-supplied founder STs, and can optionally correct IC labels using a Newick tree.

    Input is a TSV of genome, ST and allele columns (e.g., FastMLST output). Clustering defaults to single-locus variants (SLV) and can be set as high as six allelic differences. Founder-ST and optional extra-ST mapping TSVs are supported. Correction acts on STs, so every isolate with a numeric ST gets a consistent IC. The only dependencies are pandas, BioPython and NetworkX. · Theory and implementation

  2. ShownIn A. baumannii, CC-only assignment inflated IC3 because a straggly CC chained in unrelated STs, including the ATCC 17978 laboratory strain, and the phylogeny showed these isolates as non-cohesive.

    ST437 (ATCC 17978), which to the authors' knowledge had never been assigned to IC3, was placed in IC3 by SLV clustering of 1,281 chromosomes. In the Fig. 1 inner ring, isolates labelled IC3 are scattered widely across the core-genome phylogeny, so the excess reflects aberrant CC linkage rather than real evolutionary structure. · Assignment of Acinetobacter baumannii isolates…; Figure legends

  3. SuggestedTree-based correction with a 'Not IC' subtree threshold of 2 made IC3 coherent and kept the other ICs. The default threshold of zero removed probable true IC members.

    Some monophyletic IC groups contained 'Not IC' isolates, so threshold 0 dropped likely valid IC isolates. Threshold 2 on the IQtree v2.4.0 core-genome tree of 1,281 A. baumannii chromosomes kept IC3 coherent and the other ICs intact (Fig. 1 outer ring). Per-IC counts appear in Table 1, which is not reproduced in the text. · Theory and implementation; Assignment of Acinetobacter baumannii isolates…; Figure legends

  4. ShownAt the SLV level, every IC fell within one CC except IC7 and IC3. Each split because a literature-assigned ST was two alleles from the founder with no intermediate ST, and the ST-mapping input recovered both.

    IC7: founder ST25 vs ST113. IC3: founder ST3 vs ST229. Each pair differs at 2 alleles in this 1,281-chromosome dataset. ST113 and ST229 were assigned to IC7 and IC3 through the known-ST mapping file, and their strains sat close to the respective founder STs on the phylogeny. · Assignment of Acinetobacter baumannii isolates…; Tables

  5. ClaimedClustering at the double-locus-variant (DLV) level would merge the split ICs but caused spurious grouping of unrelated STs. This motivated keeping SLV as the default and relying on phylogenetic correction.

    The Discussion reports this without counts or a table. The SLV default split IC7 (ST25/ST113) and IC3 (ST3/ST229) in the A. baumannii set. · Discussion; Assignment of Acinetobacter baumannii isolates…

Methods, data and reproducibility

Allelic profiles are linked into clonal complexes when they differ at no more than a user-set number of loci (default SLV). Complexes that contain user-specified founder STs are labelled as ICs. Optionally, labels are reassigned to tree subtrees that hold a founder ST and no more than a threshold number of 'Not IC' isolates.

  • Previously assembled set of A. baumannii chromosomes, typed with the Pasteur MLST scheme and used for CC clustering and core-genome phylogeny · Acinetobacter baumannii · n = 1,281 chromosomes
  • Acinetobacter nosocomialis chromosome used as the outgroup to root the core-genome tree · Acinetobacter nosocomialis · GenBank CP157432.1
  • Founder STs and additional IC-associated STs compiled from published literature, used to map CCs and STs to ICs · Acinetobacter baumannii
Methods & tools
PhyloMLST (Python 3 CLI), FastMLST v0.0.16 (Pasteur MLST scheme), bakta v1.11 (light database), Panaroo v1.5.2, MAFFT, IQtree v2.4.0, ModelFinder, TreeViewer, pandas, BioPython, NetworkX
Code
https://github.com/Mattn286/PhyloMLST
Data & accessions
https://github.com/Mattn286/PhyloMLST
Software & environment
PhyloMLST runs on Python 3 with pandas, BioPython and NetworkX (versions not stated). The analysis used FastMLST v0.0.16, bakta v1.11 (light database), Panaroo v1.5.2 with MAFFT (core threshold 1), and IQtree v2.4.0 with ModelFinder.
Reproducibility gaps
Accessions for the 1,281 A. baumannii chromosomes are not listed; the set is described only as previously constructed. PhyloMLST dependency versions are not pinned, and the per-IC isolate counts in Table 1 are missing from the text.
Limitations (5)
  • Correction needs an accurate phylogeny built from every isolate in the dataset. Recombination could distort the tree, although the authors argue that a topology-only method is little affected and that recombination-free trees can be used instead. · stated by the authors
  • In testing, the default 'Not IC' subtree threshold of zero was often too strict because of DLVs and dropped likely valid IC isolates. Users can optionally raise it; 2 was used for A. baumannii. · stated by the authors
  • SLV-level clustering split literature-defined ICs (IC7, IC3) into separate CCs, while DLV clustering produced spurious merges. Correct IC membership therefore depends on user-supplied ST mappings or phylogenetic correction. · stated by the authors
  • Validation used a single A. baumannii dataset with no ground-truth labels and no quantitative comparison with PubMLST, goeBURST/PHYLOViZ or the recently published founder SLV/DLV Python workflow. The threshold of 2 was also chosen on that same dataset. · noted in this summary
  • Isolates with non-numeric STs (novel allelic profiles) were masked, so handling of novel profiles, a weakness attributed to goeBURST, is not demonstrated. Use with other species such as K. pneumoniae clonal groups is asserted but untested. · noted in this summary

Who should read:Read this if you type bacterial isolates by MLST and need a scriptable way to assign STs to clonal complexes or named epidemic lineages such as A. baumannii international clones.

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

Similarity Map
SelectedSame fieldCross-field
THISPalma '26Venditti '18Ovsepian '26Ryan '26Hennart '21Huebner '20
67% semantic matchMicrobiology
Féderica Palma, Mélanie Hennart, Keith A. Jolley+19 more·PLoS Biology·Jun 4, 2026· 19 citations