A functional assay measures what a variant does in an experimental system. That can be highly informative, especially when clinical observations are sparse. Yet a measurable molecular difference is not automatically a disease-causing difference. The experiment must connect the observed effect to the relevant disease mechanism and distinguish clinically meaningful disruption from tolerated variation.

Illustrated overview of from an assay to interpretation, highlighting measured function, clinical relevance, evidence calibration.
From an assay to interpretation: measured function, clinical relevance, evidence calibration. Read the article for evidence and limitations. Original conceptual illustration.

Define what the assay measures

ClinGen’s recommendations for PS3 and BS3 describe a structured approach to evaluating functional evidence. They emphasize the disease mechanism, the relevance of the assay and its validation with appropriate controls. The evidential weight should reflect those properties rather than the sophistication of the technology or the apparent size of an experimental effect. Read the functional-evidence recommendations.

A normal result can be particularly difficult to interpret. An assay of one protein function may miss effects on expression, splicing, localization or another function. Its ability to exclude disease mechanisms should be demonstrated rather than inferred from the absence of a difference in one readout.

Scale does not replace calibration

Findlay and colleagues’ 2018 BRCA1 saturation genome-editing study showed how thousands of variants can be assessed in their genomic context. The work is an important example of high-throughput functional characterization, while remaining tied to particular exons, domains, cell biology and measured outcomes. Results from that assay should not be generalized automatically to all BRCA1 variants or other genes. Read the saturation-editing study.

For a new assay, select control variants independently of the assay result and avoid circular validation. If classifications used as the reference already depend on the same experiment, agreement can exaggerate performance. Replication, batch behavior and intermediate results deserve attention alongside the separation of clearly benign and clearly pathogenic controls.

Report evidence that can be reused

  • Describe the biological function measured and the mechanisms it cannot assess.
  • Provide calibration data and uncertainty around the result.
  • Explain how assay output maps to an evidence strength.
  • Avoid treating multiple correlated readouts as independent proof.

The research conclusion and the clinical conclusion should remain distinguishable. A variant can be functionally abnormal without sufficient evidence for a clinical classification, and a disease-associated variant can appear normal in an incomplete model. Good functional studies make those boundaries explicit so their findings can be integrated responsibly with population, family and clinical evidence.