The ΔΔCt method for qPCR, with a worked example
ΔΔCt is the most common way to turn qPCR Cq values into a fold change in gene expression. It is simple arithmetic, which is exactly why it is so easy to get subtly wrong in a spreadsheet.
The ΔΔCt method calculates relative expression as fold change = 2^(−ΔΔCt). First normalise each sample to a reference gene: ΔCt = Cq(target) − Cq(reference). Then compare to the calibrator (usually the control group): ΔΔCt = ΔCt(sample) − ΔCt(calibrator). For example, if treated cells have ΔCt 6.0 and control cells 8.3, ΔΔCt is −2.3 and the fold change is 2^2.3 ≈ 4.9, so the gene is about five times higher in treated cells. The method assumes both assays amplify with close to 100% efficiency.
The formula
Three steps, all on Cq values (the cycle at which fluorescence crosses the threshold):
- Average replicates for each sample and gene.
- ΔCt = Cq(target gene) − Cq(reference gene), for each sample.
- ΔΔCt = ΔCt(sample) − ΔCt(calibrator), then fold change = 2−ΔΔCt.
The base 2 comes from the assumption that the product doubles every cycle, i.e. 100% efficiency.
A worked example
| Target Cq | Reference Cq | ΔCt | |
|---|---|---|---|
| Control (calibrator) | 26.5 | 18.2 | 8.3 |
| Treated | 24.0 | 18.0 | 6.0 |
ΔΔCt = 6.0 − 8.3 = −2.3.
Fold change = 22.3 ≈ 4.92. The target gene is about 4.9-fold higher in the treated sample. A positive ΔΔCt would mean lower expression: ΔΔCt = +1 gives 0.5, half the control level.
Assumptions to check before you trust the number
- Efficiency close to 100% for both target and reference, and similar to each other. Check with a standard curve; if efficiencies differ substantially, an efficiency-corrected model such as Pfaffl's is more appropriate.
- A stable reference gene under your conditions. Using two or more reference genes and their geometric mean is more robust than one.
- Specific amplification: a single melt peak, and clean no-template controls.
- Consistent threshold and baseline per target across all plates you compare.
Statistics: do them on ΔCt, not fold change
Fold changes are not normally distributed; a 2-fold increase and a 2-fold decrease are not symmetrical around 1. Run t-tests or ANOVA on ΔCt (or ΔΔCt) values from biological replicates, then convert means and confidence intervals to fold change for the figure. Report biological, not technical, replicates as n.
Doing it straight from the run file
In a spreadsheet, every step above is a chance for a copy-paste slip. In CycleView you open the run file, mark the reference genes and the calibrator group, and the ΔCt, ΔΔCt, fold change with error bars and statistics are computed from the Cq values in the file. Move a threshold and everything downstream updates. Export the chart as SVG for the paper.
Open your run on your own laptop
CycleView opens QuantStudio, LightCycler, Rotor-Gene and RDML files on Mac and Windows. The viewer is free, and nothing is uploaded.
Get CycleView
Mac and Windows · runs locally · for research use, not for diagnosis.
FAQ
Is ΔΔCt the same as 2^-ΔΔCt or the Livak method?
Yes. The comparative Ct method, the Livak method and 2^-ΔΔCt all refer to the same calculation.
Can I use more than one reference gene?
Yes, and it is recommended. Use the mean of their Cq values (equivalent to the geometric mean of their quantities) as the reference.
What if my efficiencies are not 100%?
If both are similar and within the usual 90–110% range, ΔΔCt is generally acceptable. If they differ a lot, use an efficiency-corrected calculation.
Should n be technical or biological replicates?
Biological replicates. Technical replicates are averaged first; they measure pipetting precision, not biological variation.