From the editor Original
01
Welcome
Reading the field
About ninety papers a week match the query behind this feed, and seven in ten of the 23,000 records it matches have appeared since 2021. What a site like this can do about that is narrow, and worth being precise about: make the literature legible, and say plainly what a small number of papers found — and what they did not.
Curated studies Editor's selection
The Lancet Microbe · Jun 2026
Rapid diagnosis of common, undetected, and uncultivable bloodstream infections from positive blood cultures using Oxford Nanopore sequencing
211 positive blood cultures at Oxford University Hospitals, sequenced direct from the bottle and compared against routine culture and MALDI-ToF. Species identification reached 97% sensitivity and 94% specificity at a median 3 h 20 min — roughly 10 hours ahead of routine reporting — and surfaced 19 infections the standard workflow had not identified, 13 of them polymicrobial.
Clinical bottom lineThe largest direct-from-bottle ONT dataset published to date, and its own conclusion is the useful one: identification is ready well before resistance interpretation is. AMR results landed 20 h ahead of phenotypic AST but at 88% overall sensitivity, ranging from 100% for S. aureus down to 91% for E. coli. The authors call for multisite validation before clinical implementation.
Microbiology Spectrum · Feb 2024
Rapid nanopore sequencing and predictive susceptibility testing of positive blood cultures from intensive care patients with sepsis
Prospective enrolment across four intensive care units, sequencing human-depleted DNA from positive blood culture broths on MinION, with cloud-based in silico susceptibility prediction from assembled contigs. Species agreement was 94.2%, rising to 100% in monomicrobial infections, and results were reportable 8–16 h from bottle positivity.
Clinical bottom lineThe error rates are the part worth reading. Categorical agreement of 89.3% reads acceptably until you look at direction: very major errors — false-susceptible calls — ran at 12.1% on high-quality predictions and 28.4% once exploratory calls were included, and performance was poor for P. aeruginosa. In sepsis that is the dangerous direction to be wrong in, and it is the clearest published case for reporting resistance determinants rather than inferred susceptibility.
Microbiology Spectrum · Aug 2026
Random PCR-based nanopore whole-genome sequencing enables pre-positivity detection of fungal bloodstream infections
Rather than waiting for the bottle to flag, this workflow samples blood cultures drawn before positivity, depletes host DNA by saponin lysis and benzonase, then amplifies and sequences on MinION with real-time taxonomic classification. Across 48 clinical samples it gave species-level identification within 1.5 h of starting sequencing, around 7 h end to end, and resolved both fungal–fungal and fungal–bacterial co-infections.
Clinical bottom lineTwo things here matter beyond the headline. It moves the clock back from “positive bottle” to “bottle in the incubator”, and species calls needed as few as 4,000 reads — which changes the economics of real-time streaming, not just the speed. Note the claim is candidate resistance-associated variants, not antifungal susceptibility.
Transboundary and Emerging Diseases · Aug 2026
Companion animals harbour globally circulating human-associated Klebsiella pneumoniae lineages and high-risk antimicrobial resistance clones
A global assembly of 712 dog- and cat-derived K. pneumoniae genomes from 25 countries, compared against 38,106 human-associated isolates. The animal isolates spanned 263 sequence types but kept returning to the same high-risk lineages — ST307, ST11, ST15, ST147. ESBL genes were present in 42.9% and carbapenemase genes in 13.9%, and 71.1% of companion-animal sequence types were also found in people.
Clinical bottom lineThe detail that matters is where the resistance was found: multidrug resistance was not confined to isolates taken from sick animals, which makes carriage in a healthy pet a reservoir rather than a curiosity. Cat-derived isolates were more often multidrug-resistant than dog-derived ones (80.0% against 56.3%), largely through enrichment of ST147. A study of this shape is only possible at this scale, and the scale is the argument for sequencing outside the hospital.
Journal of Clinical Microbiology · Aug 2026
Applicability of Nanopore-only whole-genome sequencing for Pseudomonas aeruginosa outbreak investigation in the ICU setting: a multicentric study
Nineteen P. aeruginosa isolates from Brussels intensive care units, already sequenced on Illumina, were re-sequenced with a Nanopore-only workflow on V14 chemistry and put through two analysis platforms. Both agreed closely with the Illumina data; Ridom Typer came within a maximum of one cgMLST allele and confirmed every previously identified transmission cluster.
Clinical bottom lineThis is the question a laboratory actually has to answer before dropping short reads for outbreak work, asked directly rather than by inference: can long reads alone reproduce the clusters the reference method found? Here they could, with the caveat that it is nineteen isolates from one centre's collection. Note also that the analysis platform mattered as much as the chemistry — the two pipelines did not perform identically on the same reads.
BMC Bioinformatics · Aug 2026
plsMD: a plasmid reconstruction tool from short-read assemblies
Plasmids fragment in short-read assemblies because they are full of repeats, which leaves resistance genes catalogued but not placed. plsMD guides reconstruction using replicon and full-plasmid reference databases on top of Unicycler assemblies. On an established benchmark it reached 91.3% recall, 95.5% precision and an F1 of 92.0%, outperforming existing tools; on a freshly sequenced set the same figures were 77.6%, 88.9% and 74.5%.
Clinical bottom lineWorth reading against the drop from benchmark to new data, which is the more honest number and still the best of the tools compared. Knowing that a resistance gene sits on a conjugative plasmid rather than the chromosome is what separates a strain that is resistant from a strain that can make its neighbours resistant, and short-read surveillance archives mostly cannot tell you which. Tools that recover that context retrospectively are doing more than tidying assemblies.
Microbiology Spectrum · Sep 2026
Comparative evaluation of probe-capture and conventional metagenomic sequencing across multiple clinical sample types, with analysis of paired bronchoalveolar lavage fluid and blood samples
Host DNA swamps conventional metagenomic sequencing in low-biomass samples. This compared probe-capture enrichment against conventional mNGS on 282 clinical specimens — lavage fluid, blood, CSF — then looked at 621 paired lavage-and-blood samples from sepsis patients with pulmonary infection. Probe capture detected pathogens in 66.67% of samples against 57.10%, and the gain was largest in blood: 66.67% against 47.52%.
Clinical bottom lineThe enrichment gain is real and it is biggest exactly where diagnostic sequencing struggles most, which is blood. But the concordance figures are the ones to take to a validation plan: lavage and CSF agreed well between methods (96.30% and 88% overall) while blood did not, at 70.92% overall and 54.05% negative agreement. Two methods disagreeing on half the negatives in blood is a specimen problem before it is a platform problem.
Clinical Microbiology Reviews · Aug 2026
Highly pathogenic avian influenza viruses in mammals: host-range expansion and implications of milk-borne viral shedding
A review tracing H5N1 from the 1878 description of fowl plague to the continued global circulation of clade 2.3.4.4b, built around the molecular features that let the virus cross species — the polybasic cleavage site in haemagglutinin, and the “1+7” ribonucleoprotein arrangement of a single polymerase core with seven RNA segments. Its pivot is 2024, when H5N1 appeared in US dairy cattle.
Clinical bottom lineThat detection was the first confirmed instance of a mammalian host shedding the virus into milk, which puts a transmission route on the table that the avian reservoir model never had to account for. It is a review rather than new data, and it earns its place for framing rather than for figures: the host-range question has moved from whether the virus can infect mammals to what a mammalian reservoir would mean for surveillance that is still pointed mostly at birds.
Nature Medicine · Aug 2026
Emergence and spread of Plasmodium falciparum PX1 polymorphisms associated with decreased susceptibility to antimalarials in Uganda
A selection scan across 157 P. falciparum whole genomes from Uganda. Extended haplotypes did appear around the known Kelch13 C469Y and A675V mutations — but the strongest signal of recent selection sat somewhere else entirely, on a stretch of chromosome 7 encoding a phosphoinositide-binding protein, PX1. The haplotype was first seen in 2008 and passed 50% prevalence in northern Uganda by 2016 and eastern Uganda by 2023.
Clinical bottom lineThe authors did not stop at the association. Parasites carrying the haplotype were measurably less susceptible ex vivo to lumefantrine, mefloquine and dihydroartemisinin, and disrupting px1 in vitro made a strain more susceptible to all three — genotype, phenotype and a functional test in the same paper. It is also a case against surveillance built only on known markers: a Kelch13 panel would have registered nothing unusual here, because the locus under strongest selection was not on the panel.
Microbiology Spectrum · Sep 2026
Rapid ESBL-Carba Combo PN test for simultaneous detection of ESBL and carbapenemase activity in Enterobacterales
A chromogenic assay that detects extended-spectrum β-lactamase and carbapenemase activity in one workflow, and separates the carbapenemase classes by adding relebactam as a selective class A inhibitor. It comes from the University of Fribourg and NARA, the Swiss national reference centre for emerging antibiotic resistance, with a co-author at the Belgian diagnostics manufacturer D-Tek. The isolates used to validate it had resistance mechanisms already defined by whole-genome sequencing: ESBL detection reached 97.8% sensitivity at one hour (n = 67) and carbapenemase activity 100% at twenty minutes (n = 92), both at 100% specificity, for under a dollar a sample and about three minutes of hands-on time.
Clinical bottom lineNot a rival to sequencing; the phenotypic confirmation a genotypic result asks for, and note that the paper's own reference method was whole-genome sequencing. Gene presence and enzyme activity are different measurands, and for carbapenemase you now need both: a negative gene panel cannot exclude carbapenem resistance arising from porin loss with AmpC hyperproduction, and an activity assay reports a class rather than a gene. Class D is still where phenotypic methods are weakest, at 91.3% sensitivity after twenty minutes. Neither approach replaces full susceptibility testing; neither produces an MIC.
About
WGS Pulse tracks how genomic sequencing is changing diagnostic microbiology — pathogen identification, resistance genotyping, outbreak investigation, and the laboratory workflows around them. Articles are written by and for the people who run the assays and act on the results: laboratory scientists, clinicians, infection-control teams and bioinformaticians.
The live feed is automated: one PubMed query applied identically to every result, not hand-picked. The query, and what has changed in it, is available on request. WGS Pulse is edited, and curated items selected, by Will Hendy, Biomedical Scientist (HCPC registered), specialising in microbiology.
DisclosureThe editor is a co-founder of BaseTrack Limited, which develops diagnostic sequencing software and workflows for clinical microbiology. WGS Pulse is not a BaseTrack product, and covers the field rather than the company.