For researchers
From microbiome study design to structured analytical outputs.
Study scoping, specimen workflows, sequencing, reproducible bioinformatics, governed cohort architecture and ATLAS structured outputs — for principal investigators, research groups and clinical research teams.
Where VAMS fits in a study
- Study scoping Question, population, comparison group and what the microbiome layer can realistically contribute.
- Specimen workflow Collection, handling and chain of custody designed to survive an audit rather than an inspection.
- Sequencing 16S amplicon, shotgun metagenomics or expression-context work, chosen against what the question needs.
- Bioinformatics Reproducible, versioned pipelines — a result from March remains comparable with one from October.
- Reference & cohort context Governed cohort architecture with explicit provenance, so a comparison group is a stated choice.
- ATLAS outputs Scores, panels and Insight Packs with confidence and evidence context attached to each output.
- Longitudinal analysis Repeat timepoints on pinned pipeline versions with deterministic build manifests.
- Structured reporting Machine-readable outputs alongside human-readable reports.
The anatomy of a reproducible cohort study
A study is an architecture before it is a dataset: the question is fixed first, the sampling matrix carries the design, and the provenance record underneath is what makes a re-run comparable.
01 · Research question
A question fixed before collection begins — cohort, comparison, timeframe and endpoint defined up front, so the design is answerable rather than exploratory after the fact.
02 · Participants
Strata defined, consent scoped
Eligibility, arm allocation and the consent scope for secondary use are recorded before the first sample is issued.
03 · Metadata schema
The same fields, every visit
A fixed schema captured at each timepoint. Fields added mid-study create gaps that cannot be filled retrospectively.
04 · Longitudinal sampling
Repeat sampling on a fixed cadence. A missed timepoint is recorded as missing, never back-filled.
05 · Sequencing
Batched with controls
Runs carry negative and positive controls; batch identity travels with every sample.
06 · Analysis
Pinned pipeline, pinned parameters
Analysis runs against a versioned pipeline and a recorded parameter set, so the same inputs give the same output.
07 · Reproducibility — the provenance record
- Participant ID (pseudonymised)
- Arm and timepoint
- Collection metadata
- Kit and batch lot
- Protocol version
- Extraction and run ID
- Pipeline version
- Parameter set
Reproducibility is the point
A microbiome result that cannot be reconstructed is not a research asset.
- Accepted input Raw reads or already-processed tables, with the method and run details that produced them. FASTQ readsProcessed tables
- QC and method fit Read quality, depth, completeness and metadata are checked, and the data is tested against what the method can support. Does not pass: held and returned with the QC reasons, rather than analysed with a caveat.
- Analysis Taxonomic or functional assignment, run on a recorded pipeline version.
- Structured output Tables and figures returned with the method, pipeline version and the limits that apply to them.
Versioned pipelines
Every analytical artifact carries its pipeline version, so a result can be reconstructed rather than approximated.
Deterministic manifests
The same frozen inputs and versions produce the same manifest hash — comparability is verified, not assumed.
Cohort provenance
Reference cohorts are immutable once approved; changing membership creates a new version rather than editing history.
Graded evidence
Five evidence tiers gate what an output may say, with the strongest tier requiring replication across independent datasets.
Stronger support
- Hidden from consumer output by default
Emerging
Governed interpretation
Tier decides how strongly a statement may be made, and whether it may be made at all. It is a measure of support in the literature — not certainty about a person.
Research areas we can scope
- Gastrointestinal research
- Inflammatory bowel disease
- Metabolic health
- Liver and metabolic disease
- Healthy and reference populations
- Longitudinal microbiome programmes
- Women's-health microbiome research
- Oral–systemic research
Research and collaboration areas, not diagnostic indications. Analytical validity is not clinical utility, and neither establishes causation.
Before you write to us
We have not decided on a sequencing approach yet. Can we still talk?
Yes. Method choice should follow the research question, not precede it. Every technical field on the enquiry form accepts "Not sure yet".
We already have FASTQ data. Can you work with it?
Often, yes. The workflow accepts existing data — raw FASTQ or processed tables — from the laboratory you already use, so you need not change your sequencing arrangements to collaborate.
Who owns the data and any resulting IP?
Proposed as a starting point: background IP stays with its owner, jointly generated IP is negotiated before work begins rather than after results appear, and publication rights are agreed up front. Nothing is settled until it is written down.
Can VAMS support an ethics submission?
We can supply method, workflow and data-handling documentation for your submission. Ethics approval itself sits with your institution.
Is any of this validated for clinical use?
No. This is research collaboration. Analytical output is not clinical validation, and research use is not validated clinical use.
Tell us about your study
Early-stage projects are welcome. Every technical question below accepts “Not sure yet” — method choice should follow the research question, not precede it.
Thank you. Your enquiry has been received.
Our team will review your organisation, goals and requested pathway, and route the enquiry to the appropriate VAMS team.