Getting started
Ampliova is organized around batches. A batch is an experiment: it groups one or more plates that share the same calibrator group and analysis method. When samples are too numerous for a single plate, multiple plates can be combined, provided an inter-run calibrator (IRC) sample is present on every plate to enable cross-plate normalization. All plates in a batch must include the reference genes.
Create a batch
From the dashboard, click New batch. Give your experiment a name, declare your target genes and reference genes, and select the calibrator group (the control condition your fold changes will be relative to). You can also choose between efficiency-corrected quantification and classic ΔΔCq at this stage.
Add a plate
Upload an RDML file exported from your instrument. Ampliova parses the file, extracts Cq values and raw amplification curves, and pre-fills the well layout from the sample and target names already embedded in the file. Review the layout and assign sample names, group names, and well types (sample, reference, NTC). Click on any well to inspect its amplification curve. When everything looks right, click Validate plate.
Run analysis
Once at least one plate is validated, the Run analysis button becomes active in the Plates tab. Click it: the engine runs LinRegPCR on all wells, applies Grubbs outlier exclusion, computes efficiencies and fold changes, and runs the appropriate statistical test automatically. Results appear in the Results tab within seconds.
The science
Why efficiency-corrected?
The classic ΔΔCq method (Livak and Schmittgen, 2001) assumes that every gene amplifies with exactly E = 2.0, meaning 100% efficiency, perfect doubling each cycle. In practice, this assumption is almost never met. Primer design, template quality, inhibitors in the reaction, and even pipetting variation all affect efficiency.
When the efficiency of your target gene differs from your reference gene by even 5%, the ΔΔCq method introduces a systematic bias that compounds with every cycle. At Cq 25, a 5% efficiency difference produces a ~30% error in the reported fold change.
Ampliova measures efficiency directly from the exponential phase of each amplification curve using the LinRegPCR algorithm (Untergasser et al. 2021). No standard curves are required. No dilution series. The efficiency is computed from the data you already have, per reaction, and then averaged per gene across all replicates in the batch, discarding outlier wells via Grubbs test first.
This approach is the analytical standard required by MIQE 2.0 (Bustin et al. 2025), the updated minimum information guidelines for publication of qPCR experiments.
Complete workflow
Here is what happens under the hood when you click Run analysis:
Every intermediate value (per-well efficiency, Rq, Rq_ref, Nq, calibrator mean) is stored and accessible in the Table view and the JSON / XLSX export. Full traceability from raw Cq to fold change, in a single file.
Formulas
All quantities are computed per sample per target gene.
For the classic ΔΔCq method, E = 2.0 is used for all genes and the LinRegPCR step is skipped.
File formats
Ampliova accepts RDML files (.rdml), the open standard for qPCR data interchange, supported by all major instrument manufacturers. RDML files include raw amplification curves, which are required for LinRegPCR efficiency estimation.
RDML is supported natively by Bio-Rad CFX, Applied Biosystems QuantStudio, Roche LightCycler, and QIAGEN Rotor-Gene Q, among others. See the export instructions below for your instrument.
Using QIAGEN RT² Profiler arrays? See the RT² Profiler guide →
How to export RDML
QuantStudio (Applied Biosystems)
- Open your experiment in Design & Analysis Software.
- Go to File → Export, select RDML as the format.
- Make sure Include amplification data is checked; this is required for efficiency calculation.
Bio-Rad CFX Maestro
- Open your run in CFX Maestro.
- Go to File → Export → RDML.
- Select Include raw fluorescence data and export. The
.rdmlfile is a ZIP archive; upload it directly; Ampliova unpacks it automatically.
Interpreting results
Fold change and SE
Fold change is the expression of a sample relative to the calibrator group. By construction, the mean fold change of the calibrator group equals 1.0. A fold change of 2.0 means the target gene is expressed at twice the level of the calibrator; 0.5 means half.
The standard error (SE) is propagated analytically through the full calculation chain: from per-replicate Rq values, through reference gene normalization, to the final fold change. The error bars in the Figures tab represent ±SE. This is not the SD of fold changes across replicates; it is a true propagated uncertainty that accounts for technical variation at every step.
A note on scale: fold changes are on a linear scale. A change from 0.5× to 2.0× represents a 4-fold difference in expression, not 1.5×. For visualizing large dynamic ranges, consider using ΔΔCq (also exported) which is on a log₂ scale.
Statistical tests
Ampliova selects the statistical test automatically based on the number of groups, normality, and variance homogeneity. Tests are performed on fold change values across biological replicates.
- Normality is assessed by Shapiro-Wilk on each group.
- Variance homogeneity is assessed by Levene's test.
- A warningis shown when n < 5 in any group: the Shapiro-Wilk test is underpowered at small n and normality cannot be reliably determined.
Reference gene stability
After each analysis, Ampliova evaluates the stability of your reference genes using four complementary methods and recommends the most stable one.
geNorm M
The average pairwise variation of a reference gene with all other reference genes, computed as the standard deviation of log₂ ratios across all samples (Vandesompele 2002). Threshold: M < 0.5.
BestKeeper MAD
Mean Absolute Deviation of raw Cq values across all samples: MAD = mean(|Cq − mean(Cq)|). Implemented as in RefSeeker (Asmussen & Dalgaard 2024, Bio-Protocol). Threshold: MAD < 0.5.
ΔCq SD
The mean standard deviation of pairwise Cq differences (Cq_i − Cq_j) across all samples (Silver et al. 2006). A stable gene maintains a constant Cq difference relative to all others.
NormFinder
Computes per-sample deviations from the geometric mean of all candidates, then estimates inter-group and intra-group variation of those deviations (Andersen et al. 2004). Lower is better. Threshold: score < 0.15.
Ampliova computes all four scores, ranks genes per method, then combines them using the geometric mean of ranks, the RefFinder algorithm (Xie et al. 2012), as implemented in the RefSeeker R package (Asmussen & Dalgaard 2024, Bio-Protocol). The most stable gene is highlighted in the Results tab.
Best pair for normalization
Once individual genes are ranked, Ampliova selects the best pair to use as your normalization reference: since MIQE 2.0 requires ≥ 2 reference genes, the pair matters more than the individual rank.
Primary method · NormFinder pair: all possible gene pairs are evaluated by computing their combined NormFinder stability score (inter-group + intra-group variation of the pair together). The pair with the lowest combined score is recommended. This is the most sensitive method when your samples span distinct biological groups (e.g. WT vs. KO).
Fallback · geNorm pair: if NormFinder pair scoring is not available (fewer than 2 groups, or single sample), the two genes with the lowest combined rank (RefFinder) are selected. Their geNorm M score is the standard deviation of log₂(Cq_gene1 / Cq_gene2) across all samples.
Example output · 3 reference genes
| Gene | geNorm M | BestKeeper MAD | ΔCq SD | NormFinder | Combined rank |
|---|---|---|---|---|---|
| GAPDH | 0.38 | 0.42 | 0.31 | 0.08 | ★ 1Most stable |
| RPL13a | 0.47 | 0.68 | 0.44 | 0.11 | 2 |
| ACTB | 0.71 | 1.23 | 0.82 | 0.21 ⚠ | 3 ⚠ |
⚠ Values above thresholds (geNorm M > 0.5, BestKeeper MAD > 0.5, NormFinder ≥ 0.15) suggest unstable expression. Consider replacing this gene.
FAQ
How many reference genes are required?
At least 2 are strongly recommended (MIQE 2.0, Bustin 2025; Vandesompele 2002). A single reference gene cannot be validated for stability. Ampliova displays a warning if only one gene is declared.
My efficiency is 1.77. Is that too low?
No. Ampliova estimates efficiency with LinRegPCR, directly from the amplification curves, and LinRegPCR values run systematically lower than standard-curve values. The expected window is 1.70 to 1.90 (Ruijter et al. 2009). The familiar 90 to 110 % criterion from MIQE (E ≈ 1.8 to 2.2) describes a dilution-series efficiency, so applying it to a LinRegPCR estimate is the most common way these numbers get misread. If a value does fall outside 1.70 to 1.90, inspect the amplification curve: a poor baseline, late exponential phase, or flat curve indicates a technical problem. Outlier wells are excluded automatically by the Grubbs test (p < 0.05). Systematic out-of-range efficiency across an assay suggests a primer or reagent issue.
Can I analyze multiple plates in a single batch?
Yes. That is the central architecture of Ampliova. A Batch groups any number of plates. Mean efficiency is computed across all replicates in the entire batch, which increases statistical power and reduces plate-to-plate variation.
What is the difference between Rq and Nq?
Rq (relative quantity) = E^(−Cq), the raw efficiency-corrected quantity for a single gene in a single sample. Nq (normalized quantity) = Rq_target / Rq_ref, after dividing by the geometric mean of the reference genes. Fold change is computed from Nq.
Is my data confidential?
Yes. All data (including Cq values and amplification curves) are stored on European servers (Supabase, EU West/Ireland; Vercel EU) and never leave the EU. For details on sub-processors and data retention, see the Privacy Policy.
Is Ampliova citable in a publication?
Yes, and you should. In your Methods section, name Ampliova as the software used (e.g. "qPCR data were analyzed using Ampliova, ampliova.com"), and cite Müller and Tiret (2026, Bio-Protocol), "Efficiency-Corrected Relative Quantification of qPCR Data Using LinRegPCR and a Spreadsheet-Based Workflow", as the primary methodological reference. Also cite the underlying methods as appropriate to your analysis: LinRegPCR (Untergasser et al. 2021, BMC Bioinformatics); the ΔΔCq method if used instead (Livak and Schmittgen, 2001, Methods); geometric mean normalization (Vandesompele et al. 2002, Genome Biology); reference gene stability: NormFinder (Andersen et al. 2004, Cancer Research) and/or BestKeeper (Pfaffl et al. 2004, Biotechnology Letters) if reported; and MIQE 2.0 (Bustin et al. 2025, Clin Chem). The auto-generated Methods section (Export → Methods) includes the Ampliova and primary-reference citations automatically.
What does the MIQE 2.0 badge mean?
Ampliova follows the Minimum Information for Publication of Quantitative PCR Experiments guidelines, version 2.0 (Bustin et al. 2025, Clin Chem) by default. This includes: mean efficiency per gene via LinRegPCR (averaged across all replicates in the batch), geometric mean normalization across ≥2 reference genes, outlier exclusion by Grubbs test, appropriate statistical test selection, and full audit trail export.
Can I use the classic ΔΔCq method instead?
Yes. When creating or editing a batch, select "Classic ΔΔCq (Livak 2001)" as the method. Ampliova will assume E = 2.0 for all genes and skip the LinRegPCR step. The rest of the pipeline (normalization, calibration, statistics) is identical.
Scientific references
Ampliova's analytical pipeline is grounded in peer-reviewed methodology. If you use Ampliova in a publication, please cite the relevant papers below.