Frequently asked questions

At least two (Vandesompele et al. 2002; MIQE 2.0, Bustin et al. 2025). A single reference gene cannot be checked for stability, since there is nothing to compare it with, and Ampliova warns when a plate carries only one. A validation batch tells you which pair to use.

No. Ampliova estimates efficiency with LinRegPCR, from the amplification curves, and LinRegPCR values run lower than standard-curve values: 1.70 to 1.90 is the expected range. The familiar rule of 90 to 110% (E between 1.90 and 2.10) is a convention for dilution series, not a MIQE threshold: MIQE sets no range and asks for the confidence interval. A gene outside 1.70 to 1.90 deserves a look at its curves: a poor baseline, a late or flat exponential phase, or a primer problem. Single wells that do not fit their gene are removed by a Grubbs test before the mean is taken.

Yes. Type the wells of the series as standards (Std) with their dilutions, on their own plates in a Primer efficiency batch or next to your samples, and Ampliova fits one standard curve per gene: E = 10^(−1/slope), with its 95% interval, R² and linear range. In the Efficiencies panel of a batch you then choose, gene by gene, LinRegPCR, the curve on that batch, or an efficiency saved in your library, including one measured elsewhere and typed in with its method. The source is named next to every value, in the exports and in the Methods paragraph. Fitting curves is free; saving and reusing efficiencies is part of the Researcher plan.

They never enter the calculation: they are checked. Each control is compared with the samples it guards on the same plate, and one that comes up less than 5 cycles after its sample (about 3% of its signal) is flagged when you add the plate and in the MIQE checklist. Name each RT− after the RNA it was made from ("WT_1 RT-"), since genomic DNA contamination differs from one preparation to the next.

Yes. A batch pools any number of plates: each gene's efficiency comes from all its wells, and a sample's replicates are averaged wherever they sit. Keep all samples of a gene on the same plate if you can (sample maximization), and plate effects cancel out of every comparison. If a gene's samples have to be split across plates, turn on inter-run calibration and run the same calibrator for that gene on every plate: each plate is then shifted onto a common scale before the analysis, as qbase does.

Rq, the relative quantity, is E^(−mean Cq) for one gene in one sample. Nq, the normalized quantity, is the Rq of the target divided by the geometric mean of the Rq of the reference genes. The fold change is Nq divided by the mean Nq of the calibrator group.

It depends on the design, never on the data. Two groups: Welch's t-test. Three groups and up: an ordinary one-way ANOVA followed by Dunnett's test against the calibrator, or Tukey's test for every pair. Paired designs: the paired t-test, or the repeated-measures ANOVA with the Geisser-Greenhouse correction. Two-factor designs: a two-way ANOVA with Type III sums of squares. All run on log2(fold change), and every comparison is reported as a ratio with its 95% confidence interval. Alternatives are in the Assumptions tab.

Welch's test does not assume that the two groups share a variance, and when they do, with equal group sizes, it loses almost nothing. It is the one default that is right either way. Student's test remains available.

Because Ampliova tests log2(fold change), where the multiplicative noise of qPCR becomes additive and the groups share a spread. A t-test on raw fold changes violates its own assumptions. Run Welch's test on log2 of the fold changes (in Excel, T.TEST with type 3 on the LOG of the values) and you get Ampliova's P value.

Something makes the test impossible, and the reason is shown under the gene: a group with a single sample, values that do not vary, a pair ID used twice in the same group, or a two-factor design with an empty combination of factors. Ampliova reports it rather than dropping the gene silently.

A power computed from the effect you just observed is a function of the P value and adds nothing to it: a non-significant result always has low power to detect its own effect (Hoenig and Heisey 2001). The confidence interval under the chart shows which effects the data rule out and which they cannot, which is what that power figure is usually meant to convey.

Yes. The Prism export holds, for each gene, a table of fold changes and the same table on log2. Run Prism's unpaired t test with Welch's correction, or its ordinary one-way ANOVA with Dunnett's or Tukey's test, on the log2 table and you get the same P values. Ampliova's own results are checked against R, which Prism also reproduces.

Yes. Choose "Classic ΔΔCq (Livak 2001)" when creating or editing the batch. Every gene then gets E = 2, LinRegPCR and the Grubbs test are skipped, and the calibrator is averaged on the Cq scale, as Livak and Schmittgen define it. Normalization and statistics are the same.

Cq values, amplification curves and results are stored in Supabase's EU West region (Ireland), and the analysis engine runs in the EU (Railway, Netherlands). Plate-layout suggestions send sample and gene names, never personal data, to Mistral AI in Paris. Each user can turn them off in Settings, and a lab's owners and admins can turn them off for every member; nothing is sent then. Every sub-processor and its region is listed in the Privacy Policy.

Yes. Name the software in your Methods ("qPCR data were analysed with Ampliova, ampliova.com") and cite Müller and Tiret (2026, Bio-protocol) as the method it implements, together with LinRegPCR, geometric-mean normalization and MIQE 2.0. The Methods button in the Results tab writes a paragraph with these citations from your batch.

It lists the MIQE 2.0 items the analysis step can show, and says which your batch meets: efficiency correction, at least two reference genes, their stability, controls, specificity, replicates, statistics. Compliance with the guideline depends on the whole experiment, from sample handling to primers, so no software can grant it; the checklist tells you where your batch stands on the part Ampliova sees.