qPCR Relative Quantification
Open the Cq export from your instrument. Replicates are averaged and checked, each gene is corrected for its efficiency, targets are normalised to the geometric mean of your reference genes, and fold changes come out with error bars done on the log scale — every step shown.
1 · LOAD SECTION 02 — CALCULATORS · 2.5
2 · DESIGN GROUPS · GENES · EFFICIENCY
3 · CHECK
4 · RESULTS FOLD CHANGE = 2 ^ MEAN log₂ NRQ
| GENE | GROUP | n | FOLD CHANGE | SEM | log₂ ± | p (Holm) |
|---|
EACH BIOLOGICAL REPLICATE tap a row for the working
| SAMPLE | GROUP | GENE | MEAN Cq | log₂ RQ | log₂ NF | log₂ NRQ | NRQ |
|---|
How to use
- Export the results table (Cq per well) from the instrument software — see the table below — and open it here. Several plates or one file per gene are fine.
- Check the groups: one row per biological sample, same group name for replicates of a condition. Mark the reference genes.
- Set the efficiency per gene if you have measured it. Standard-curve wells in the file are picked up and fitted.
- In step 3, look at the flagged replicate sets. Tap a Cq to leave that well out — nothing is removed for you.
- Read the fold changes, then take the chart (PNG/SVG), the CSV with every intermediate value and the methods paragraph. SAVE RECORD keeps the wells and your choices so you can reopen the analysis later.
The formula
| RQ | E ^ (Cqcalibrator − Cqsample) for each gene |
| E | 10 ^ (−1 ÷ slope) of the standard curve; 2 = 100% |
| fold change | 2 ^ mean(log₂ NRQ) of the biological replicates |
| error bar | 2 ^ (mean ± SEM), on the log₂ scale |
Example. With E = 2: target Cq 25.0 and reference 18.0 in treated cells, 27.0 and 18.0 in the control. log₂ RQ target = 27 − 25 = 2, reference = 18 − 18 = 0, log₂ NRQ = 2 → 4-fold. With a target efficiency of 1.90 instead: log₂ 1.90 × 2 = 1.852 → 3.61-fold.
Good to know
- Free. Nothing you open leaves your browser; the last analysis stays on this device until you press CLEAR.
- Statistics are on biological replicates only. Technical wells tell you how well you pipetted, not how variable the biology is.
- The tool can't judge melt curves, amplification plots or primer specificity — check those in the instrument software.
Exporting from your instrument
Open the results table (one Cq per well), not the run file. Menu names change between software versions; these are the usual routes.
| Instrument | Export | What the tool reads |
|---|---|---|
| Bio-Rad CFX (Maestro, CFX Manager) | Export → all data sheets, then open the “Quantification Cq Results” file (.csv or .xlsx) | Well, Target (or Fluor), Content (Unkn / Std / NTC / NRT), Sample, Biological Set Name, Cq, Starting Quantity |
| Thermo QuantStudio, ViiA 7, 7500, StepOne | Export → Results, as .xlsx or .txt (an old .xls: save it as .xlsx first) | Well Position, Omit, Sample Name, Target Name or Detector, Task, CT / Cq, Quantity — the [Results] section of a text export |
| Roche LightCycler 480 | Abs Quant results table → export as text; one file per channel | Include, Pos, Name, Cp, Standard. There is no gene column — name the gene for each file after opening |
| Roche LightCycler 96 | Export the analysis results to Excel | Well, Sample Name, Gene Name, Cq |
| Qiagen Rotor-Gene Q | Quantitation results → export to Excel or .csv | No., Name, Type, Ct, Given Conc |
| Any instrument | RDML (.rdml) from the instrument or RDML tools; or paste from Excel | Sample, target, Cq per reaction; sample type and standard quantity |
Not on the list, or the columns aren't found? Tell us the instrument and we'll add it — we'll ask for an example export by email. Old binary .xls and run files (.eds, .pcrd, .ixo) can't be read — export the results table instead.
Checks the tool runs
| Flag | When | What to do |
|---|---|---|
| HIGH SD | Replicate SD above your limit (0.3 cycles by default) | Look at the wells. For triplicates the tool points to the well furthest from the median when dropping it brings the SD below 0.2. |
| NO AMP / NOT DETECTED | Undetermined wells | Left out of the mean, never set to 40. A gene not detected in a sample is not calculated for it. |
| LATE Cq | Mean Cq above 35 (adjustable) | Low copy number: expect more scatter; consider more template. |
| EFFICIENCY / R² / FEW POINTS | Standard curve outside 90–110%, R² below 0.98, fewer than 5 dilution levels | Check the dilutions and primer design; MIQE asks for a curve over at least 5 points. |
| geNorm M | Two or more reference genes | Below 0.5 is stable for homogeneous samples, below 1 acceptable for tissue. |
| REFERENCE SHIFT | Reference level differs by more than 0.5 log₂ units between groups | The treatment may change your reference genes — validate them. |
| NTC AMPLIFIED | NTC or no-RT well within 5 cycles of the latest sample | Contamination or genomic DNA; primer dimers show in the melt curve. |
LAST UPDATED · HOW VALUES ARE CHECKED · ABOUT LABMATE.TOOLS
Frequently asked
How do I calculate fold change with the ΔΔCt method?
For each sample, ΔCt = Ct(target gene) − Ct(reference gene). ΔΔCt = ΔCt(sample) − ΔCt(calibrator, usually the mean of the control group). Fold change = 2−ΔΔCt. Example: target Ct 25.0 and reference 18.0 in treated cells (ΔCt 7.0), 27.0 and 18.0 in control (ΔCt 9.0): ΔΔCt = −2.0, fold change = 4.
When should I use the Pfaffl method instead of 2−ΔΔCt?
2−ΔΔCt assumes every assay doubles its product each cycle (efficiency 100%, E = 2). If your standard curves give efficiencies that differ from 100% or from each other, use the efficiency-corrected (Pfaffl) calculation: RQ = E(Cq calibrator − Cq sample) for each gene, then target RQ ÷ reference RQ. With E = 2 for every gene the two give the same answer.
How do I normalise to more than one reference gene?
Divide the target's relative quantity by the geometric mean of the reference genes' relative quantities (Vandesompele 2002, Hellemans 2007). If the efficiencies differ, averaging the reference Ct values first is not the same thing — compute each reference's efficiency-corrected quantity, then take the geometric mean.
How should I calculate error bars for qPCR fold change?
Work on the log scale. Compute log₂ relative expression (−ΔΔCq) for each biological replicate, then take the mean and SD or SEM of those values. The fold change is 2mean and the error bar runs from 2mean − SEM to 2mean + SEM, so it is not symmetric. Averaging fold changes directly, or putting the SD of the fold changes on the bar, overstates up-regulation and understates down-regulation.
Which calibrator should I use?
Usually the mean of the control group. Then the control group's geometric mean is exactly 1 but each control replicate keeps its own value, so the control bar has an error bar too. Picking a single control sample as the calibrator makes every result depend on that one sample.
What should I do with a replicate well that is far from the others?
Look at it before you drop it. Technical replicates usually agree within 0.2–0.3 cycles. The tool flags sets with SD above your limit and, for triplicates, points to the well furthest from the median when dropping it brings the SD below 0.2. It never removes a well by itself; the wells you exclude are listed in the CSV and the methods text.
What does “Undetermined” mean, and should I set it to 40?
Undetermined means the well never crossed the threshold. Replacing it with 40 invents a number. The tool leaves such wells out of the mean and flags the set; if a gene is undetermined in every well of a sample, that sample is not calculated for that gene.
Is my data uploaded?
No. Files are read and calculated in your browser and are not sent anywhere. The last analysis is kept on this device until you press CLEAR.