How most thesis qPCR is analyzed today
A template someone in the lab built years ago: Cq values pasted from the cycler, ΔΔCq against GAPDH, the fold change copied into Prism for the figure. It is how many labs teach qPCR, and for a first result it works.
The questions come back later. Across published studies the average is 1.2 reference genes, and 15% tested their candidates for stability first (Chapman and Waldenström, 2015). Normalizing on one unvalidated gene was off by up to threefold in a quarter of the cases tested by Vandesompele et al. (2002).
“Even after about 20 years of complaining, we still see people using reference genes ‘just because’. And readers have no way to check that the reference gene was stable enough to detect the signal. Seriously, in practice nobody uses two, let alone three!”
| Today | With Clarida |
|---|---|
| GAPDH, because the lab has always used it | geNorm and NormFinder rank your candidate reference genes in the experiment, and normalizing on one gene shows a warning |
| One efficiency of 100% for every assay | Efficiency per target from your dilution series, with its error carried to the result |
| Cq values pasted from the cycler’s export | Cq read from the run file, or called from the raw curves |
| Months later, the reviewer asks how you analyzed it | The report has already written the methods sentences from your settings |
On the lab’s Excel template? When is your spreadsheet enough?
The four questions a reviewer asks
MIQE 2.0 (Bustin et al., 2025) recommends against plain ΔΔCq. It asks for reference genes justified by their stability, efficiency-corrected quantities and the raw data, shared as RDML or RDES. Clarida analyzes your run that way, so each question a reviewer raises has its answer in the experiment.
- Which reference genes, validated how? Your candidates are scored for stability in the experiment itself, and you choose which to keep (reference genes). Both also run as free tools, with no account: geNorm and NormFinder.
- At what efficiency? Per target, from your dilution series. Its error is carried into the fold change, with the spread of your replicates. See Analyze.
- How did you analyze it? The report writes its methods sentences from the settings you used: the reference genes, the test, the groups and the correction (Report).
- Can we see the data? Export RDML with the raw curves, for the data availability statement a journal asks for.
Start from the run you already have
- Your run file. CFX, LightCycler 480, QuantStudio, Rotor-Gene or RDML, curves included. Or a Cq table from any instrument.
- Your instrument’s Cq stays. It is kept beside the Cq Clarida calls from the curves, and you pick which one feeds the result (importing instrument data).
- Work half done in Excel or Prism. Bring the Cq values and go on from there; Prism stays for the figure.
- qbase+ archives. They import free, read-only and outside your experiment count.
For your PI
A PI’s questions before saying yes, each with its answer.
| What it costs | About €1,800 a year for a lab running 10 experiments a month, at the academic rate of €15 each, with no charge per seat (pricing). |
|---|---|
| Where the data lives | In the EU: the app runs in Amsterdam, the database in Stockholm, and everything travels over HTTPS. |
| Who sees it | The members of your lab’s workspace. Anyone else needs an invitation, which comes with the paid plans. |
| Getting it out | Each experiment as RDML, an Excel workbook or its report, at any time. |
| If you stop paying | Your experiments stay viewable and exportable, but you cannot create new ones. |
| The next student | Experiments belong to the lab’s workspace, not to a person, and reopen as they were left. For published papers, the odds that the data still exist fall by 17% a year (Vines et al., 2014). |
| Grants and journals | RDML is the open data format for qPCR: one to name in a data management plan. |
| Who built it | From the people behind qBase, qbase+ and geNorm. Jan Hellemans led the qBase paper, Jo Vandesompele the geNorm one, and both co-wrote the MIQE guidelines (meet them). |
| Not yet | Two-way ANOVA (use Prism for it), absolute quantification, and exporting all experiments at once. |
Frequently asked questions
Do you need your PI’s approval to start?
No. The free plan is yours alone: no card, nothing to install. Your PI comes in when the lab wants a shared workspace.
Half your data is in Excel or Prism. Do you redo it?
No. Import the Cq values you already have and carry on in Clarida.
Can you reproduce the numbers later?
Yes. A saved report is frozen, and a change becomes its next version. The History tab keeps earlier states of the experiment’s design and settings too, to compare and restore.
Is unpublished data safe?
It is hosted in the EU and sent over HTTPS, and it stays in your lab’s workspace, which others join only by invitation. The security notes have the details.
Can your PI check the analysis?
Yes, on a paid plan: invite them into the workspace. The free plan is for one person.