Methodology in detail

What we measure from a single sample.

BAARK profiles bacteria, fungi and selected parasites from one faecal sample, and quantifies a defined set of oral bacteria from one swab, all processed in our Melbourne laboratory. The seven cards below jump to each part of the test.

Bacteria

The gut community, sequenced and counted.

Two measurements from one faecal sample. Shotgun metagenomic sequencing on Oxford Nanopore reads the whole bacterial community. A quantitative PCR (qPCR) panel counts the core bacteria that published veterinary studies rely on.

What we measure

The whole community. The composition and relative abundance of the bacteria in your pet's gut. Shotgun sequencing reads every microbial DNA fragment in the sample, not a single marker gene, so coverage runs from the most abundant groups down to low-abundance taxa that smaller assays miss.

The core taxa, counted. Eight qPCR assays measure total bacteria and seven groups: Faecalibacterium, Turicibacter, Streptococcus, Escherichia coli, Blautia, Fusobacterium and Clostridium hiranonis. These are the groups used in the canine dysbiosis index published by the Gastrointestinal Laboratory at Texas A&M University [1].

How we measure it

We extract total DNA from a small faecal sample once, then split it.

  • Sequencing. Most of the DNA is sequenced on Oxford Nanopore and the reads are classified against the GTDB r214 reference database. The output is a taxonomic table resolved to species, and to strain where an organism is covered deeply enough.
  • qPCR. An aliquot runs on a QuantStudio 1. Seven of the eight assays use the published primer sequences unchanged [1]. The published Streptococcus primers matched only a small fraction of Streptococcus reference sequences in our check, so we use a redesigned genus-level pair. Each assay is quantified against a synthetic DNA standard curve, so the result is an absolute quantity, not a share of reads.

Why both. Sequencing shows what is there, including organisms no panel targets. qPCR measures a fixed set precisely and comparably from one test to the next. Some core taxa, such as C. hiranonis, can go undetected by sequencing, while targeted qPCR detects them with higher analytical sensitivity [2].

Dogs and cats. The sequencing workflow is the same for both. The core-taxa panel and its published reference ranges come from dogs; cats have their own published panel built on a different set of bacteria [3].

What it tells you

What makes up your pet's gut community, and where each core taxon sits as an absolute quantity. The report flags organisms of interest, summarises diversity, and lists organisms worth discussing with your vet. Comparison with other pets is in development, as the BAARK reference cohort is built.

What it does not tell you

The presence of a particular bacterium is not a diagnosis. The same organism can be harmless in one animal and contribute to problems in another, depending on community context, host genetics, diet and medication history. The report describes what we found; clinical interpretation belongs with your vet.

Diet moves these numbers in healthy dogs. In a 2025 study, dogs fed fresh (lightly cooked) diets had more Streptococcus and E. coli, and dogs fed kibble had more Fusobacterium and C. hiranonis; average scores stayed in the normal range on every diet [2]. A shift in a fresh-fed dog may reflect what it eats rather than a problem.

Oral microbiome

A second window.

An oral swab adds a second sample type, measured a different way. The gut is sequenced to find out what is present. The mouth is quantified by targeted qPCR to find out how much, across 8 bacteria studied in canine periodontal disease.

What we measure

The bacteria in your pet's mouth, sampled by swabbing the gumline and inner cheek. Like the gut, the oral community shifts with age, diet and dental health. It is a different ecosystem from the gut, under different pressures, and changes there can show up before gut changes do.

The species involved in canine periodontal disease are named in the published literature, so an open-ended read of the whole community is not what this sample needs. A defined panel measured accurately is more useful than a broad panel measured loosely.

How we measure it

Quantitative PCR, not sequencing. An aliquot of the swab eluate runs across an 8-target panel plus a universal bacterial 16S reaction, on the same QuantStudio 1 used for the parasite screen. Every target is quantified against a synthetic DNA standard curve, so the result is an absolute quantity for that species rather than a share of a read pool. The universal reaction gives total bacterial load, which each target is normalised against.

Oral qPCR panel
TargetCanine Oral TaxonGroup
Treponema denticolaDisease-associated
Porphyromonas gulaeCOT-052Disease-associated
Tannerella forsythiaCOT-023Disease-associated
Fusobacterium nucleatum
Prevotella intermedia
Campylobacter rectusCOT-010 lineage
Parvimonas micraCOT-076
Porphyromonas cangingivalisCOT-109
Universal bacterial 16S rRNAn/aNormaliser for total bacterial load

The COT numbers are Canine Oral Taxon designations. Only about 16 per cent of canine oral taxa are shared with the human mouth [5], so assays built on human strains do not always transfer. Our C. rectus assay, for example, matches the canine oral lineage, not the human type strain. Where a COT number is listed, the assay was checked against that canine organism.

Sample collection: oral swab, included in the Bundle kit or available as the Oral Add-on for existing customers. If the panel changes, this page changes first, with the date.

What it tells you

How much of each target is present in your pet's mouth, as a quantity rather than a percentage, normalised against total bacterial load.

Because every figure is an absolute quantity measured against a standard curve, it does not move just because something else in the sample moved. A relative abundance does: a species can appear to rise only because another fell away. That makes an absolute number the better measurement.

It does not make a difference between two tests automatically meaningful. No published study reports within-dog variation in an untreated animal, so the size of change that counts as real is not yet known. We report the number and the direction; we do not claim a threshold for meaningful change.

What it does not tell you

The oral panel is not a substitute for a dental examination by a veterinarian. We measure what is in the sample; we do not assess plaque, tartar, pocket depth, tooth wear or the structure of the teeth and gums. A high result is a flag for veterinary review, not a diagnosis.

A targeted panel measures its targets and nothing else. Eight species is not the whole oral microbiome, and a species outside the panel will not be reported, however abundant it is.

The periodontal risk score is in development. It has not been calibrated against animals with vet-confirmed periodontal staging, and we claim no sensitivity or specificity for it.

The published canine evidence is still small. In a 2025 study of 52 dogs, T. denticola had the strongest link to periodontal disease severity, while P. gulae on its own showed no significant link [4]. Nothing published supports a diagnostic accuracy claim for a combined oral panel in dogs.

Cats. The assays were checked against canine oral bacteria. We report the same quantities for cats, but no published study links them to dental disease in cats.

Fungi

Captured by the same sequencing.

Shotgun NGS reads every DNA fragment in the sample, not just bacterial DNA. Fungal reads are picked up alongside, and clinically relevant species are flagged when present.

What we measure

Fungal organisms present in the sample, identified by their genomic DNA picked up during shotgun metagenomic sequencing. Panel-based commercial tests detect a defined list of fungal species; because BAARK sequences untargeted, fungal DNA appears in the read pool whenever it is present at detectable abundance, including species not on any commercial panel.

How we measure it

No separate workflow. Reads are taxonomically classified across kingdoms, so fungal hits emerge from the same sequencing run as bacterial hits. The pipeline uses fungal reference databases to assign reads to genera and species where the evidence supports it.

  • Malassezia, Candida, Aspergillus and other clinically relevant species are flagged when present at detectable abundance.
  • Fungal sensitivity depends on relative abundance in the sample; very low-abundance fungi may not produce enough reads for confident species-level assignment.

What it tells you

Whether named fungi of clinical interest were detected in the sample and at what relative abundance. This adds context to skin, ear, or gut concerns where fungal involvement is plausible from the published veterinary literature.

What it does not tell you

The detection of Malassezia or Candida in a sample does not, on its own, mean your pet has a fungal infection. These genera include both commensal and pathogenic species, and the clinical relevance of any specific finding depends on the animal's symptoms and history. The report informs a veterinary conversation; it does not replace one.

Parasites

A targeted screen, alongside the broad one.

Six gut parasites, screened by quantitative PCR on the same faecal sample, in addition to anything shotgun sequencing picks up.

What we measure

The presence and relative quantity of DNA from six gut parasites. Shotgun sequencing catches parasites when they are abundant; targeted qPCR adds sensitivity for low-level infections that sequencing might miss.

How we measure it

The faecal DNA is split: most goes to sequencing, and an aliquot runs through the qPCR panel on a QuantStudio 1. Each reaction carries an amplification control, and each sample carries an extraction control, so a negative result means the target was not detected rather than that the reaction failed.

Parasite qPCR panel
TargetCommon name
Giardia duodenalisGiardia
Cryptosporidium spp.Cryptosporidium
Trichuris spp.Whipworm
Strongyloides spp.Threadworm
Ancylostoma brazilienseHookworm
Uncinaria stenocephalaHookworm

In development: Ancylostoma caninum (hookworm). It is not reported until its assay is validated.

What it tells you

Whether DNA from any of the six target parasites was detected in the sample, and at what relative quantity. A positive result is a flag for veterinary review and possible follow-up testing. A negative result means the six targets were below the detection threshold in this sample.

What it does not tell you

The panel covers six targets, not every parasite that can affect a pet. It does not test for roundworm (Toxocara), coccidia (Cystoisospora) or flea tapeworm (Dipylidium caninum); ask your vet about a faecal float if those are a concern. A negative result does not rule out other parasites. A positive result shows parasite DNA was present, but does not on its own establish active infection, its severity or treatment need; that is a veterinary judgement.

BAARK Biome Health Engine

From sequencing reads to a readable report.

Sequencing produces millions of reads, and qPCR produces exact quantities for the core taxa. The BAARK Biome Health Engine turns them into an interim gut health score, organism-of-interest flags, and diet and supplement points for you and your vet to discuss. Our own dysbiosis index is in development, built from the BAARK reference cohort.

What we measure

Three things are computed at this stage. First, an interim gut health score, calculated from the relative abundance of species on a curated list of beneficial and concerning gut bacteria. It is not yet calibrated against a reference population or validated against animals with a known diagnosis. Our own dysbiosis index is in development, built from the BAARK reference cohort. Second, organism-of-interest flags for species on curated lists of pathogens and other organisms worth discussing with your vet. Third, diet and supplement points to discuss with your vet, drawn from published veterinary microbiome research.

How we measure it

Bioinformatics first. Read classification against GTDB r214 and the qPCR standard curves do the heavy lifting. The Engine then applies rule-based scoring developed by our scientific team from the published veterinary literature.

  • Recommendations are generated per sample, not pulled from a generic list.
  • The Engine is updated as the published literature and our reference cohort grow.

How our score relates to the published dysbiosis index. The canine dysbiosis index uses the same eight qPCR assays as our core panel [1]. It was built to tell healthy dogs from dogs with biopsy-confirmed chronic enteropathy; at its cut-off of 0 it identified 74 per cent of affected dogs and 95 per cent of healthy dogs [1]. Our interim score is not that index and is not calculated the same way, so the two numbers cannot be swapped for each other. We will not report a BAARK dysbiosis index until our own, built from the BAARK reference cohort, has been validated.

What it tells you

A summary you can read in five minutes and bring to a vet appointment. The interim score gives a single high-level read; the organism-of-interest flags give specifics; the recommendations give starting points for dietary and supplement discussion.

What it does not tell you

The Biome Health Engine produces wellness profiling, not clinical diagnosis. The interim score summarises the species we found; it is not a measure of disease severity and has not been validated against animals with a known diagnosis. Recommendations are intended to inform a veterinary conversation, not to replace one. Always consult a veterinarian before starting, stopping, or changing any treatment for your pet.

One sample. Five analyses.

The same faecal sample feeds the bacterial profile, the core-taxa qPCR panel, the fungal screen, the parasite qPCR and the BAARK Biome Health Engine report. The Bundle adds an oral swab, measured on its own qPCR panel.

The DNA is the DNA. Once it is in the lab, we can ask many questions of it, rather than making one sample do the work of one fixed assay.

  1. Sample arrives in Melbourne Stabilised in transit, processed in our laboratory.
  2. DNA extraction and split Most of the extract goes to sequencing; aliquots go to the qPCR panels.
  3. Sequencing and qPCR run in parallel Shotgun sequencing on Oxford Nanopore. Core-taxa and parasite qPCR on QuantStudio 1.
  4. Bioinformatics and Engine Reads are classified against GTDB r214, qPCR results are read off their standard curves, and the Engine builds your report.
  5. Results in your portal 7 to 10 business days from sample receipt.
Longitudinal tracking

Trajectory beats snapshot.

Microbiome composition responds to diet, antibiotics, illness, and short-term stress. Repeat testing over months or years carries more signal than any single result.

What it tells you

Whether your pet's microbiome is moving in the direction you and your vet are aiming for, after a diet change, antibiotic course, probiotic supplementation, or a deliberate weight or wellness intervention. The trajectory is the answer to "is this working?"

How it works

Order another Gut Health or Bundle test on the cadence that suits your pet (typically 3, 6, or 12 months). Each sample is processed exactly the same way, so results are directly comparable. The portal shows the change over time alongside the absolute values.

What it does not tell you

Microbiome shifts are influenced by many factors at once: diet, environment, season, stress, medications. A change between two timepoints does not, on its own, attribute the change to any specific intervention. Discuss interpretation with your vet, especially if a course of treatment is involved.

Research access

PetTrials, included free.

Every BAARK customer gets free access to PetTrials, an Australian clinical trial matching platform for veterinary research, run by our sister company.

What you get

A PetTrials account that matches your pet to active veterinary research studies based on species, age, breed, and (with your permission) the health context you shared with us at order. Studies range from nutrition trials and probiotic intervention research to specific-condition cohorts.

How participation works

Matching is opt-in: you receive notifications about studies your pet might qualify for, and you decide whether to enrol. PetTrials is a separate platform; BAARK shares only the metadata you authorise.

What BAARK is, and what it isn't.

BAARK is a wellness profiling service. It is not a diagnostic test, and it is not intended to diagnose, prevent, monitor, or treat any disease. The reports describe what we find in the sample; clinical interpretation, treatment decisions, and follow-up belong with your veterinarian.

If your pet is showing acute symptoms (severe vomiting, blood in stool, weight loss, lethargy, dehydration), see a vet directly. BAARK is designed for proactive wellness profiling, post-treatment recovery tracking, and longitudinal monitoring of changes in your pet's microbiome, not for emergency care.

References

  1. AlShawaqfeh, M. K., Wajid, B., Minamoto, Y., Markel, M., Lidbury, J. A., Steiner, J. M., Serpedin, E., & Suchodolski, J. S. (2017). A dysbiosis index to assess microbial changes in fecal samples of dogs with chronic inflammatory enteropathy. FEMS Microbiology Ecology, 93(11), fix136. doi.org/10.1093/femsec/fix136
  2. Oba, P. M., Roberts, L. J., Geary, E. L., Suchodolski, J. S., & Swanson, K. S. (2025). Effects of diet type on the core fecal bacterial taxa and the dysbiosis index of healthy adult dogs. Frontiers in Veterinary Science, 1572875. doi.org/10.3389/fvets.2025.1572875
  3. Sung, C.-H., Marsilio, S., Chow, B., Zornow, K. A., Slovak, J. E., Pilla, R., Lidbury, J. A., Steiner, J. M., Park, S. Y., Hong, M.-P., Hill, S. L., & Suchodolski, J. S. (2022). Dysbiosis index to evaluate the fecal microbiota in healthy cats and cats with chronic enteropathies. Journal of Feline Medicine and Surgery, 24(6), e1-e12. doi.org/10.1177/1098612X221077876
  4. Takahashi, K., Arima, E., Sawayama, H., Suda, A., Miura, N., Shirakata, Y., Ito, S., & Suda, Y. (2025). Possible correlation of specific bacteria detected by using a novel screening method and periodontal disease in dogs. Journal of Veterinary Medical Science, 87(6), 580-588. doi.org/10.1292/jvms.25-0004
  5. Dewhirst, F. E., Klein, E. A., Thompson, E. C., Blanton, J. M., Chen, T., Milella, L., Buckley, C. M. F., Davis, I. J., Bennett, M.-L., & Marshall-Jones, Z. V. (2012). The canine oral microbiome. PLoS ONE, 7(4), e36067. doi.org/10.1371/journal.pone.0036067