Skip to content
VAMS LABS™ · quality & validation

Quality controls and validation boundaries, shown clearly.

Credible microbiome work depends on honest quality practice — and on being clear about what “validation” does and does not mean. This page describes how the pipeline is checked and where its limits lie.

Ask About Validation See the Sample Workflow
Sample integrity checksExtraction & run QCBatch / contamination awareDocumented workflowMethod-bounded interpretationResearch-use framing
Validation philosophy

Method validation vs clinical validation — kept distinct

Method/workflow validation means a workflow reliably does what it is designed to do for research-use outputs. Clinical validation is a separate, regulated process establishing performance for medical decision-making. Unless a specific clinical pathway is verified and formally approved, VAMS LABS™ describes the former, not the latter.

  • ✓Method/workflow validation — consistency and reproducibility of research-use outputs.
  • ✓Clinical validation — a distinct, separately governed and regulated process.
  • ✓Outputs are educational and non-diagnostic unless a separately approved pathway exists.

Reproducibility

How reproducibility is approached

Analytical reproducibility is assessed using replicate quality monitoring and laboratory quality-assurance procedures. Quantitative figures are established during laboratory validation and are not stated speculatively.

Replicate quality monitoring across runs and batches
Versioned pipelines keep analyses comparable over time
Laboratory quality-assurance procedures applied throughout
Batch and contamination effects considered in design and analysis

Sequencing quality & coverage

Depth, coverage and quality filters

Sequencing depth Minimum sequencing depth is established according to the validated laboratory protocol.
Coverage Coverage is matched to the method and research question; greater depth is not automatically more informative.
Quality filters Raw-data QC and bioinformatics quality filters are applied before any output is generated.
Reproducible pipelines Versioned, documented pipelines so results remain comparable across runs and cohorts.

Controls & quality assurance

How quality is handled — and made visible

Sample integrity & intake checks — samples are logged and reviewed at intake; integrity concerns are flagged before processing.
Extraction QC framing — extraction is assessed for quality and quantity appropriate to the chosen method.
Run-level QC — sequencing runs carry quality checks; low-quality data is identified rather than hidden.
Controls where verified — positive/negative controls are used where verified for the workflow and project scope.
Batch & contamination awareness — batch effects and contamination are considered in design and analysis, not ignored.
Re-run / review triggers — defined triggers prompt re-runs or review without exposing internal thresholds.
Documentation & audit trail — project intake, methods and versions are documented for reproducibility.
Clear limitation statements — every output states what it can and cannot support.
Detection limitations

What detection depends on

Detection capability depends on sample type, sequencing chemistry and assay configuration. Quantitative limits of detection are established during laboratory validation and are assay- and laboratory-specific — they are not stated as fixed figures until verified.

  • ✓Detection is influenced by sample type and preservation
  • ✓Sequencing chemistry and depth affect what can be resolved
  • ✓Quantitative limits are established during laboratory validation

Reference cohorts

The curated reference behind interpretation

Interpretation is contextualised against a curated reference of public datasets. These figures are traceable to the Phase 1 cohorts & datasets workbook.

Curated datasets 106 datasets across 58 curated cohorts
Expected reference samples Approximately 23,000 (sum of dataset expected counts)
Body-site systems 5 — gut, vaginal, oral, skin and multi-system
Sequencing methods 16S amplicon (79 datasets) and shotgun metagenomics (21 datasets)
Curation quality Mean admissibility score 4.4 / 5 across curated datasets

Reference-cohort details are described in full on the Reference Cohorts page (/science/reference-cohorts).

Partner laboratory attribution

Laboratory processing

Laboratory arrangements are confirmed for each implementation. VAMS BioInnovation Private Limited is not itself an accredited laboratory, and no laboratory accreditation is claimed on its behalf.

Method limitations

What this validation does not cover

Not clinical validation unless separately verified and approved Not a diagnosis, screening or treatment tool No fixed quantitative performance figures stated before verification No accreditation claimed for VAMS itself Results are method-bounded and research-use

Questions about quality or validation?

Tell us about your project and we will explain the QC practices and validation boundaries that apply to your scope.

Ask About Validation See the Sample Workflow
Hands in purple gloves hold up a 96-well assay plate with rows of pink and clear wells. Photo: James Gathany / CDC (PHIL #23208). Illustrative of laboratory practice — not a VAMS facility or team.

Looking at the evidence

Quality control means someone actually looks.

Plates, controls and checkpoints are reviewed before data moves forward. The QC gate drawn below is the formal version of this habit: inspect first, analyse second, and return what does not pass.

Quality control, illustrated

The QC gate has two outcomes.

Validation is drawn as a real gate: data that does not pass quality-control and method-fit checks is held and returned with the reasons, rather than analysed regardless.

  1. Accepted input Raw reads or already-processed tables, with the method and run details that produced them. FASTQ readsProcessed tables
  2. 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.
  3. Analysis Taxonomic or functional assignment, run on a recorded pipeline version.
  4. Structured output Tables and figures returned with the method, pipeline version and the limits that apply to them.
Illustrative analysis path with the quality-control gate emphasised: the held-and-returned branch is part of the process, not an exception.