The current discordance on Serratia spp. taxonomical diagnosis using proteomics or genomic tools.

BACKGROUND

The Serratia marcescens complex comprises several closely related species with potentially distinct epidemiological characteristics. Recent descriptions of new species further complicate accurate identification using routine clinical microbiology methods. We aimed to compare the current species-level identification by MALDI-TOF MS with Whole Genome Sequencing (WGS)-based approaches and to assess concordance among different diagnostic genomic tools.

METHODS

Overall, 107 Serratia bloodstream isolates were analysed. Initial identification was performed by MALDI-TOF MS. WGS data were analysed using: Kraken, PATO (MASH distance-based), ribosomal MLST (rMLST), and GTDB-Tk (ani-rep and classify_wf functions).

RESULTS

MALDI-TOF MS identified S. marcescens as the predominant species (71%). Kraken assigned all isolates to S. marcescens. Phylogeny-based approaches revealed a markedly different distribution. rMLST and GTDB-Tk consistently identified Serratia sarumanii as the most frequent species (36.4% vs. 34.6%), followed by Serratia ureilytica (26.2% vs. 29%) and Serratia nevei (22.4% vs. 21.5%), while S. marcescens represented a minority (5.6% vs. 3.7%). GTDB-Tk showed near-complete concordance with rMLST (95.3%). Also, 38/107 isolates assigned to S. nevei by PATO were reclassified as S. sarumanii by both rMLST and GTDB-Tk. Complete concordance across all methods was observed in only 3.7% of isolates.

CONCLUSIONS

At the moment, species-level classification of Serratia bloodstream isolates is highly method-dependent. Routine diagnostic tools tend to underestimate diversity of the S. marcescens complex, whereas phylogeny-based genomic approaches reveal a more diverse population structure dominated by S. sarumanii, S. ureilytica and S. nevei. These findings highlight the need for cautious interpretation and standardized genomic frameworks with updated databases.