A Chiba University workflow strips out human DNA, amplifies whatever is left and reads it on a nanopore — identifying eight Candida and Cryptococcus species before conventional culture declares anything at all.

A fungal bloodstream infection is a race, and the diagnostic process loses the first lap. Candida species cause most invasive fungal infections worldwide and sit among the leading causes of hospital-acquired bloodstream infection, but conventional blood culture followed by species identification takes several days. The patient is treated in the meantime on a guess.

A study in Microbiology Spectrum from Chiba University’s Medical Mycology Research Center proposes closing that gap by sequencing the blood culture bottle before anyone would normally look at it. The reported turnaround is about seven hours, and the identification happens before the culture turns positive.

The work was led by Professor Hiroki Takahashi of the Medical Mycology Research Center, with colleagues at Chiba University and the National Institute of Technology and Evaluation.

Three steps, and the first is the hard one

The workflow has three parts: deplete human DNA using benzonase, amplify whatever remains by PCR-based whole-genome amplification, and sequence it in real time on a nanopore device, checking the output against a custom reference database.

The depletion step is what makes the rest possible, and it is worth understanding why. A blood sample is overwhelmingly human — a fungal genome in that mixture is a rounding error, and sequencing it directly means reading human DNA almost exclusively. Benzonase degrades DNA that is not protected inside an intact cell, so host DNA released into the sample is destroyed while fungal cells keep theirs. What survives is enriched for the thing you are looking for.

Random PCR amplification then multiplies that remnant without needing to know in advance what it is, which is the methodological point. A targeted PCR test asks whether a specific organism is present; amplifying at random and sequencing the result asks what is present, and does not have to be told the answer first.

Takahashi described coming to the problem from outside clinical microbiology, with “a background in sequence analysis and genomics” and seeing “an opportunity to apply my expertise” to it.

Forty-eight samples, eight species

The method was tested on 48 clinical blood culture samples representing eight fungal species. Among those identified were Candida albicans, Nakaseomyces glabratus, Candida parapsilosis, Candida tropicalis and Cryptococcus neoformans.

That species list is clinically meaningful rather than incidental, because antifungal choice depends on it. N. glabratus — formerly Candida glabrata — is frequently resistant to fluconazole, the drug a clinician would otherwise reach for, and Cryptococcus neoformans is a different organism requiring different treatment entirely. Knowing which one is present on the first day rather than the third changes what the patient receives.

The release describes the accuracy as remarkably high and gives no figure for it, which is the most significant omission on the page. For a diagnostic method, the accuracy number is the result — and whether “remarkably high” means 95 per cent or 100 per cent across 48 samples determines how the work should be read.

The method also detected mixed infections in some samples, involving either two fungal species or fungi together with bacteria. Mixed fungal infections are easy to miss by culture, since one organism outgrows the other and the slower one is never reported.

What a whole genome buys

Because the approach reads the entire genome rather than a marker region, it can detect variants in drug-resistance genes alongside the species identification.

This is the part with the most headroom. Species identification narrows drug choice statistically — most isolates of a species behave a certain way — while a resistance mutation in the actual infecting organism answers the question directly. Conventional susceptibility testing requires growing the isolate and exposing it to drugs, which adds days on top of the identification. Reading the resistance genes out of the same sequencing run collapses two waits into one.

“Our method may enable clinicians to initiate appropriate antifungal treatment earlier,” Takahashi said.

He also described being “particularly motivated by the possibility of using modern genomic technologies” to improve understanding of these infections.

The limitations the authors name

Two are stated directly, and both are practical rather than conceptual.

The first is timing: further work is needed to determine the best point during incubation to sample. This is the method’s central trade-off made explicit. Sample too early and there may be too little fungal DNA to sequence; sample later and the advantage over conventional culture shrinks. Where that optimum sits — and whether it is the same for a fast-growing C. albicans and a slow-growing Cryptococcus — is unresolved.

The second is interference: detection suffers when heavy bacterial growth masks the fungal signal in mixed infections. A random-amplification approach reads whatever DNA is most abundant, and bacteria in a blood culture bottle multiply faster than fungi do. The same indiscriminate sampling that lets the method find unexpected organisms also lets an abundant one drown out a scarce one.

What it would take to use this

Several things the release does not address stand between this and a hospital laboratory.

Nanopore sequencing is relatively cheap as sequencing goes and still requires a device, reagents, and someone able to run a bioinformatic pipeline against a custom reference database at the moment a result is needed. No cost per sample is given, and no comparison against the rapid alternatives already in clinical use — such as T2Candida magnetic resonance panels or beta-D-glucan antigen testing, which are faster than culture though narrower in what they detect.

Forty-eight samples is also a feasibility cohort rather than a validation study. What it cannot establish is the false-positive rate on blood cultures from patients who turn out not to have a fungal infection, which is the number that determines whether a test like this would drive unnecessary antifungal treatment.

The funding deserves a note. The work was supported by JSPS KAKENHI grant JP22H04925, the Japan Initiative for World-leading Vaccine Research and Development Centers through Chiba University’s Synergy Center, and the Shionogi Infectious Disease Research Promotion Foundation — Shionogi being a pharmaceutical company with an infectious-disease portfolio. Foundation grants of this kind are normal and at arm’s length from a company’s commercial decisions; the release states no conflicts of interest at all, which is the detail that should have been there.

The paper was published online on 21 August 2026.

---

Notes

1. “Random PCR-based nanopore whole-genome sequencing enables pre-positivity detection of fungal bloodstream infections.” Microbiology Spectrum, 2026. DOI: 10.1128/spectrum.00478-26.
2. Chiba University. “Faster diagnosis for fungal bloodstream infections.” EurekAlert!, 5 October 2026. https://www.eurekalert.org/news-releases/1146325.

Photo by Karolina Grabowska on Pexels.com