What Effect Size Is Biologically Meaningful in Enhancer Studies?

There is no field-wide number, and papers that imply one are usually reporting a convention rather than a finding. Three anchors do the work instead:

1. Assay noise. An effect must exceed what the readout produces between replicate wells and replicate clones.
2. The smallest effect your design can detect. If your power calculation says 35%, a 25% observation is not a result.
3. The effect the biology requires. A developmental enhancer sustaining a morphogen gradient may need near-complete loss; a GWAS variant may act through a 10% change in expression and still be causal.

A defensible answer to the title question is therefore: "the smallest change that matters for the biological or clinical claim, provided it is above the assay's noise floor and within the study's power."

Two readings matter here. The bottom rows are not weaker results — they are smaller true effects. A 12% expression change from a single variant, measured consistently across clones and alleles, is a strong finding; a 30% change from a single clone is not a finding at all.

Before running anything, state the minimal effect of interest — the smallest change that would alter your interpretation of the biology. Then check feasibility:

• • Assay variability. Measure the coefficient of variation across replicate wells and across clones of the same genotype. If CV is 15%, a 10% effect is not detectable.
• • Power. With the observed variability, compute the replicate number needed to detect the minimal effect at conventional error rates. Report that number in the methods.
• • Direction. Pre-specify whether up or down, and by how much, would count. Post-hoc interpretation of direction is where most overclaiming enters.

For a screening study, the equivalent step is deciding the effect size the screen must detect, which sets library design — see our piece on how many enhancers a single CRISPRi screen can test.

Three numbers belong in every effect-size claim:

• • Relative change with a confidence interval, not just a mean.
• • Absolute change in the assay's own units, because relative changes on a low baseline can look large and matter little.
• • Context of the comparison — which control, which genotype, which cell passage.

Statements such as "reduced expression by 60%" without a control description are unverifiable, and reviewers increasingly ask for the effect size against a non-targeting control rather than against wild-type cells only.

• • Allele-specific effects. In a heterozygous background, a 15% total change can reflect a 30% effect on one allele — often the physiologically relevant number.
• • Dosage-sensitive genes. Haploinsufficient loci respond to changes that would be trivial elsewhere.
• • Program-level effects. A modest change in a single gene can shift a differentiation trajectory, which is a meaningful phenotype even if the transcript change is small.

Conversely, a large effect in one gene with no functional consequence is not automatically meaningful: effect size on a transcript is not effect size on the biology.

EDITGENE reports the performance characteristics that determine detectable effect size — replicate CV, clone-to-clone variability, dynamic range — for every engineered line, so power calculations start from measured numbers rather than literature assumptions.

Tell us the readout and the change you need to resolve. Contact EDITGENE to request a project consultation or quote.

Effect-size ranges cited here are indicative summaries of common observations, not benchmarks; values depend on locus, cell type, assay, and perturbation modality.

Biological meaning cannot be inferred from a percentage alone. Claims should connect the measured change to a functional consequence and state the assay's detectable range.

• • Fulco CP, et al. Systematic mapping of functional enhancer–promoter connections with CRISPR interference. Science. 2016;354:769–773.
• • Gasperini M, et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Cell. 2019;176:377–390.
• • Lakens D, et al. Justify your alpha. Nature Human Behaviour. 2018;2:168–171.
• • Nasser J, et al. Genome-wide enhancer maps link risk variants to disease genes. Nature. 2021;593:238–243.