In development · no release date promised
CqVerify reads your qPCR instrument's own file on your computer, recomputes Cq with three independent methods, and shows you exactly how each number was produced — which baseline, which threshold, which points.
Nothing is uploaded. There is no account. The file never leaves your machine.
In development. No delivery date until the order threshold is reached.
| Method | Cq |
|---|---|
| Single threshold | 24.31 |
| Second derivative | 24.18 |
| LinRegPCR | 24.44 |
| Spread | 0.26 |
It is that the Cq value arrives as a finished number with no visible derivation, and the software that produced it is closed. Reviewers ask how you got it. You can only restate the vendor's number. When you switch instruments, or when a colleague in another lab runs the same assay, the numbers do not agree — and there is no way to find out which one is right, or whether either is.
This is not speculation. It is in the vendors' own documentation.
“this may impact the Cq value for legacy data”
“Obsolete analysis result if the results come from an non-compatible software version.”
Opening an old experiment in a newer build can silently change your Cq. The software tells you the result is obsolete. It does not tell you the old number, or the new one.
“we cannot guarantee the performance of the software or provide support for the analyses”
…for data imported from versions the software no longer supports.
“Keep in mind that also the RDML export of the qPCR machine should contain raw, not baseline-corrected, data.”
Note the word “should”. Export shape is routinely wrong for the tool it is meant
to feed. Users hit this as list index out of bounds, or have to
re-export from the instrument by hand.
We parse the original container, so nothing is lost in an export step.
Baseline-corrected or raw? Contains negative values? Amplification count consistent with the stated chemistry? If a given Cq method cannot legally run on this file, we say so and say why — instead of returning a number or a stack trace.
This is vendor-file-level inspection. It is not an RDML conformance check.
Single-threshold · second-derivative maximum · LinRegPCR (efficiency-corrected regression), reported together with the spread between them.
If all three land within a fraction of a cycle, your conclusion is not an artifact of one threshold choice. If they diverge, you need to know before a reviewer does.
Every parameter used, every point fed to the fit, the imported vendor Cq next to each recomputed value, and an explicit statement of what the tool does not do. Built around the reporting fields MIQE 2.0 asks for.
Read and write RDML, so you can drop into the free ecosystem (rdml-python, RDML-Tools, qPCRDataMethods) for geNorm, normalisation and benchmarking. We do not reimplement those.
Read this part twice
A verification tool that overstates itself is worse than no tool, because you will trust it. So, plainly:
It will not reproduce your instrument's Cq exactly, and we will not claim it will. Vendors do not publish their baseline and threshold strategy. We guarantee reproducibility under stated parameters, not agreement with a black box.
Three methods agreeing is not proof of correctness. A 2% random baseline error alone can swing a fold-change estimate between 0.7 and 9.5. Common-mode error is real, and we do not pretend three algorithms rule it out.
No clinical interpretation. Research use only.
No experiment design, no primer design, no melt-curve analysis.
No geNorm or normalisation maths — free implementations already exist and do it well.
No cloud sync, no account, no team collaboration.
Encrypted runs are out of scope for v1.
Built for
Probably not for you if
| Instrument software | Clarida | Free R tools | CqVerify | |
|---|---|---|---|---|
| Your data leaves the machine | No | Yes (EU-hosted) | No | No — stated as a promise |
| Reads vendor file directly | Own format only | Yes (4 vendors) | No | Yes (4 vendors) |
| Shows threshold and baseline choice | No | Partial | Partial | Yes, all parameters |
| Several Cq methods together | One | Yes | One per run | Yes, with spread reported |
| Checks vendor export shape | No | RDML level (free tier) | No | Vendor-file level |
| Pricing | Free with instrument | €15 per experiment; Lab €4,800–9,600 per year | Free | €39 once |
Competitor figures checked against vendor pricing pages and release notes, October 2026. Paid tools are listed for comparison, not endorsement.
Two longer pieces, for the parts a table cannot carry:
Per-experiment billing makes you re-decide every time you analyse something, which is exactly the friction that stops people paying for software they already need. You install it; you use it.
Free
€0
No card, no expiry
Standard
€39 once
Single user · lifetime
Lab
€199 per year
Up to 5 seats
Reservations are non-binding. If we do not ship, you are not charged.
Don't take our word for it — check the basis. Our claim is narrower than “correct”. First, parsing is separated from algorithm: we publish a golden-file regression where our parse of each vendor file matches the vendor's own export, well under a tenth of a cycle, so you can see parsing errors and method errors independently. Second, our implementations are checked against the nine published methods compared in qPCRDataMethods — we report where our result falls in that distribution, not that we won. Third, every parameter of every calculation is written into the report, so the work can be redone by hand or by another tool. If a number cannot be reproduced from what we print, that is a bug worth reporting to us.
For many labs, use Clarida. It is well built, and its authors wrote geNorm and MIQE — that is not a claim we can undercut. Two reasons you might pick us instead. First, Clarida processes data on their servers; if your data is unpublished, patient-derived, or pre-patent, “your file went to a vendor's cloud” is a sentence you may not be able to write in your methods section. Second, per-experiment billing means deciding to pay every single time you analyse something. If the deciding factor is neither of those, Clarida is probably the right answer.
Not necessarily, and we will not tell you it will. Vendors do not publish their baseline-subtraction windows or threshold rules, so there is nothing to match against. What we guarantee is that under stated parameters you get the same number every time, and that the number can be traced back to specific amplification points. When our value and the instrument's differ, the report tells you which step accounts for it.
v1 targets .pcrd (Bio-Rad CFX Manager), .eds/.edt
(Applied Biosystems QuantStudio / Design & Analysis), .ixo with its
.xml sidecar (Roche LightCycler), and .rex/.qrex
(QIAGEN Rotor-Gene). Domestic Chinese instruments are a priority, not an
afterthought — none of them document RDML support, and their exports are XLS/CSV
at best.
No date, because we have not earned one. We will ship when the reservation list tells us there is something to ship for. If you need a guaranteed date today, that is a reason to buy from an established vendor instead, and we would rather you did that than waited on us.
The format parsers and algorithm implementations will be open, with a test suite and documented limitations. MIQE 2.0 asks for analysis code to be available, and a verification tool with a closed black box would be a poor advertisement for itself.
One email, at launch
Leave an email. If we ship, you hear it first, and the free tier is yours either way. If after reading the section above you think Clarida suits you better, that outcome is fine with us — we would rather be right about your situation than win the signup.
No newsletter, no resale of your address, no second email unless you reply.