The Gold Standard of Genetics | Functional Assays, Model Systems, Controls, Calibration and Variant Interpretation

THE GOLD STANDARD OF GENETICS · SPECIALIST STUDY

Treat a functional result as calibrated evidence about a mechanism, not as a verdict on a person or variant

A functional assay can show that a variant changes a measurable molecular or cellular process. That is useful only when the measured process is connected to the disease mechanism, the model can represent the relevant biology, the controls span meaningful normal and abnormal behaviour, and the result is interpreted at the strength supported by validation. A striking image or statistically significant difference is not enough. A responsible pathway defines the mechanism and endpoint before testing, separates exploratory work from a clinically validated assay, records identity and experimental provenance, handles batch effects and replicates, calibrates against independently classified variants, tests rescue or orthogonal evidence where appropriate, and keeps partial or discordant effects visible. It then integrates functional evidence with gene–disease validity, phenotype, segregation, population and computational evidence without double counting. Negative results are especially bounded: an assay that measures one function in one model cannot prove that every disease-relevant function is normal.

This guide belongs to the Genetics subject root, the Gold Standard of Education, and the How X Works hub. It begins with a concrete decision, then tests whether the method still holds when the context changes.

The whole problem at a glance

A technically precise assay can still answer the wrong biological question.

CHAPTER 2 OF 13 · Frame mechanism and question

2. Write the assay question before seeing data

Specify the endpoint, expected direction, comparison groups, decision threshold, exclusions and how a positive, negative or indeterminate result could affect evidence weight. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: The team predefines enzyme activity below a validated range as damaging support and reserves ambiguous intermediate values for no functional code. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: protocol; analysis plan; endpoint; threshold; power; exclusion criteria; evidence-strength rule; registration; deviations Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: Post-hoc thresholds can turn noise into a persuasive category, while statistical significance alone does not establish biological importance. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Move the threshold after inspecting controls and identify why the result no longer supports the original claim. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

CHAPTER 4 OF 13 · Build the model and controls

4. Choose a model system that represents context

Compare patient-derived material, engineered cells, organoids, animals and cell-free systems for tissue, isoform, developmental stage, dosage and pathway relevance. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: A neuronal splice isoform is absent from a convenient kidney cell line, so an induced neuronal model is selected despite greater complexity. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: cell or organism identity; tissue origin; differentiation; isoform; genotype; passage; culture conditions; expression level; authentication Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: Convenience and biological relevance are different properties; overexpression can create effects that endogenous dosage would not produce. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Transfer the same construct into a different tissue model and identify which conclusions require revalidation. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

CHAPTER 6 OF 13 · Calibrate and analyse

6. Build controls that span the decision range

Use wild type, null or empty controls and independently established benign and pathogenic variants that represent expected normal and abnormal behaviour. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: A multiplex assay includes several benign and pathogenic controls across domains instead of treating wild type and one severe variant as a complete calibration set. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: control selection; independent classifications; domain and mechanism coverage; numbers; blinded labels; acceptance ranges; failures Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: Circularity arises when variants classified mainly by the same assay are used to validate it, and extreme controls may not calibrate intermediate effects. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Remove benign controls and explain why a normal-looking score cannot support benign evidence reliably. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

CHAPTER 8 OF 13 · Calibrate and analyse

8. Quantify dynamic range, precision and uncertainty

Measure signal range, detection limits, repeatability, reproducibility and uncertainty around each effect rather than reporting a binary pass label alone. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: A variant score sits near the decision boundary; its confidence interval crosses the calibrated threshold, so the result remains indeterminate. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: raw values; normalisation; variance; confidence interval; limit of detection; saturation; inter-run agreement; missingness Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: A precise number can exceed the assay’s meaningful resolution, and a p-value does not show classification accuracy or effect relevance. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Keep the point estimate constant but widen inter-run variance and decide whether the evidence strength survives. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

Functional-evidence gate

The interpretation must be reproducible from the assay record and the evidence framework version.

CHAPTER 10 OF 13 · Integrate the evidence

10. Use rescue and orthogonal assays to test causality

Where appropriate, restore wild-type function, correct the variant or measure a distinct endpoint to test whether the observed phenotype follows the asserted mechanism. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: CRISPR correction normalises a cellular phenotype while an unrelated construct does not, strengthening the causal bridge without predicting patient outcome. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: rescue design; expression matching; sham editing; off-target review; independent endpoint; blinded analysis; replication; limits Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: Rescue can be non-specific or caused by overexpression, while two assays sharing the same artefact are not truly orthogonal. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Replace gene correction with high-copy overexpression and identify the new alternative explanations. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

CHAPTER 12 OF 13 · Report, review and transfer

12. Bound negative and partial results

State exactly which functions, conditions, tissues and effect sizes the assay could detect, and keep untested mechanisms visible. Begin with one variant, one asserted disease mechanism, one model system, one assay endpoint, one control set, one validation claim and one interpretation decision. State the decision, specimen, record or person, the time boundary and the responsible owner before interpreting any result.

Worked example: A normal enzyme assay in fibroblasts does not exclude a neuron-specific localisation defect or an effect visible only under metabolic stress. Put the starting evidence, applicable rule, intermediate decision, later observation and consequence on separate lines so that a fluent story cannot replace a reproducible pathway.

Evidence to retain: tested endpoint; detection capability; tissue; conditions; isoforms; power; positive-control performance; residual mechanisms Preserve provenance, dates, versions, identity checks, responsible people and intended use. If a necessary item is missing, label the gap and its effect instead of converting absence into reassurance.

Boundary and misconception: No observed effect means no effect detected under specified conditions, not proof that the variant is benign or clinically irrelevant. Test the strongest credible alternative explanation and name the new evidence that would change the conclusion. The pass condition is a bounded, reviewable claim, not the removal of uncertainty.

Evidence discipline: record the observation, version or time boundary that supports this chapter’s conclusion. Preserve the rejected alternative as well as the selected action. That makes a later review able to distinguish a changed world from an undocumented change of mind.

Changed-context transfer: Add a second disease-relevant endpoint and decide whether the combined negative evidence becomes more informative. The method transfers only if the decision question and measurement meaning survive the change. If either changes, rebuild the comparison rather than copying the earlier threshold or weight.

Contents · Previous chapter · Next chapter

Limits and misconceptions

Function is not diagnosis

An assay result is one evidence stream and must be integrated with gene validity, phenotype, inheritance and other data.

A model is a representation

Cells, organoids and animals reproduce selected mechanisms; none automatically predicts the whole person.

Normal is endpoint-specific

A negative result cannot exclude effects the assay, tissue, isoform or condition could not reveal.

Calibration does not travel automatically

Thresholds and evidence weights must be re-established when the gene, mechanism, model, platform or variant class changes.

Practise, verify and continue

Build a functional-evidence worksheet with gene–disease validity, mechanism, model, endpoint, identity, controls, replicates, batch plan, calibration, uncertainty, evidence strength, dependencies, conflicts, limits and reanalysis trigger.

Apply the method to a multiplexed assay of variant effects. Preserve mechanistic relevance, independently classified controls, calibration and uncertainty, then add variant-library representation, missing-variant handling, barcode error and saturation checks.

Practical evidence check

Before implementation, write a one-page evidence map with the decision, unit of analysis, version or time boundary, inputs, outcomes, uncertainty, authority and stop rule. Attach one worked case that passes and one that fails. A reviewer should be able to reproduce the reasoning without relying on the memory of the person who built the system.

Run one changed-context rehearsal before approval. Alter a mechanism rather than a cosmetic detail, then identify which calculation, safeguard or governance gate must change. Record any conclusion that does not transfer. This short rehearsal is often the fastest way to expose a hidden assumption before it becomes an operational dependency.

Related eduKate reading

Variant classification, evidence weight and reanalysis
Gene–disease validity and disputed genes
Genome sequencing, coverage gaps and reanalysis
Long-read sequencing and orthogonal confirmation

Authoritative sources and scope

ClinGen SVI recommendations for PS3 and BS3 functional evidence
ClinGen variant classification guidance
ClinGen general sequence variant curation process
ACMG and AMP standards for sequence variant interpretation

These sources support the laboratory and evidence distinctions described here. They do not replace patient-specific diagnostic evaluation, genetic counselling, consent, laboratory validation, local regulation or a clinician’s interpretation. All worked cases are fictional. A functional or RNA result must be integrated with phenotype, inheritance and other evidence; it must not be used alone to diagnose or change care.

Back to contents · Return to the Genetics guide