# What a qPCR script of your own asks of the lab

> When your own R or Python script is the right tool for qPCR, and what Clarida adds: the plan, the mixes and maintained methods. Each fact dated and linked.

Source: https://clarida.bio/compare/scripts
Updated: 2026-10-29

[Compared with R and Python scripts](https://clarida.bio/compare)

A script wins for methods of your own and long pipelines. Clarida covers the published ones, with the plate and mixes, without code to maintain.

[Start free](https://app.clarida.bio/login) · [Side by side](https://clarida.bio/compare/scripts#side-by-side)

4 [experiments](https://clarida.bio/pricing#whats-an-experiment) in your first month, then 2 a month. No card.

> **experiments**: The runs and analysis behind one scientific question. Counted once, when you create it. Re-analyzing, reporting and exporting are free, with no limit on samples, assays or runs. [How pricing counts experiments](https://clarida.bio/pricing#whats-an-experiment)

- **You keep**: New methods, the wider pipeline
- **You add**: The plan, the mixes, published methods

## When is your own script the better choice?

Much of qPCR analysis was first published as R and Python code, and the packages are free.

- **You are developing a new method**: New methods reach R and Python first; [qpcR](https://cran.r-project.org/package=qpcR) fits and compares amplification curve models
- **You model many genes in a complex design**: [MCMC.qpcr](https://cran.r-project.org/package=MCMC.qpcr): Bayesian models over all genes at once, with the controls as priors
- **Reference genes are one step of your pipeline**: [NormqPCR](https://bioconductor.org/packages/release/bioc/html/NormqPCR.html) has geNorm, NormFinder and ΔΔCq; [HTqPCR](https://bioconductor.org/packages/release/bioc/html/HTqPCR.html) handles high-throughput plates
- **qPCR sits next to other data**: A script runs it unattended beside sequencing, imaging or clinical data
- **A robot fills your plates**: A protocol for the [Opentrons Python API](https://docs.opentrons.com/v2/) says what goes into each well; [designit](https://cran.r-project.org/package=designit) and [Omixer](https://bioconductor.org/packages/release/bioc/html/Omixer.html) balance groups over plates
- **Someone in the lab maintains code**: With git and [renv](https://rstudio.github.io/renv/articles/renv.html), a well-kept script is as reproducible as analysis gets

## Has your script become a job of its own?

- **Only one person can run it.** When they go, the analysis leaves with them: turnover was among the difficulties people who write research software named most ([Carver 2022](https://doi.org/10.7717/peerj-cs.963)).
- **Last year’s script can stop running.** Of over 9,000 R files shared with published studies, 74% stopped with an error on a fresh setup ([Trisovic 2022](https://doi.org/10.1038/s41597-022-01143-6)).
- **An update moved the numbers.** A new major release of one imaging package changed brain volumes by 8.8% on average ([Gronenschild 2012](https://doi.org/10.1371/journal.pone.0038234)), and Bioconductor [releases every six months](https://bioconductor.org/install/), each tied to one R version.
- **A package you rely on stops.** The pcr package, [checked against published results](https://doi.org/10.7717/peerj.4473) when it appeared, was [archived on CRAN](https://cran.r-project.org/package=pcr) in November 2025 because reported issues went unfixed.
- **The plan lives somewhere else.** The plate map sits in a spreadsheet and the mix recipe on paper, so the script learns the layout only when someone types it again.
- **QA asks which wells were excluded, and why.** The answer is a line in someone’s code.

## Side by side

| What you need | R or Python scripts | Clarida |
| --- | --- | --- |
| Define |  |  |
| Starting the next experiment | **Copy and edit** Last month’s script and sample sheet, changed for the new plate | Stronger: **From a preset** Your lab’s settings, steps, assays, mix and cycling protocol at once; the goal and papers noted with it |
| Design |  |  |
| Plate layout | **Randomized** [designit](https://cran.r-project.org/package=designit) and [Omixer](https://bioconductor.org/packages/release/bioc/html/Omixer.html) spread sample groups over plates against batch effects | **Pipetting order** Laid out by your grouping and fill rules in blocks for a multichannel pipette, each automatic rule explained |
| Execute |  |  |
| Mixes and volumes | **Your own code** A few lines per mix; a package for qPCR master mixes is not documented as of October 26, 2026 | Stronger: **Worked out** What to collect and each mix recipe with its surplus, again whenever the design changes, until the runs are ready |
| Filling the plate | **Robot or paper** [Opentrons’ Python API](https://docs.opentrons.com/v2/) drives a pipetting robot; by hand, a printed plate map | **Guided** Stage by stage on the plate; a note on a well comes back next to its measurement |
| Import |  |  |
| Raw data in | **Parser needed** RDML has readers in [R](https://cran.r-project.org/package=RDML) and [Python](https://pypi.org/project/rdmlpython/); [ReadqPCR](https://bioconductor.org/packages/release/bioc/html/ReadqPCR.html) reads some instrument exports, your own code the rest | Stronger: **Run files** CFX, QuantStudio, LightCycler 480 and Rotor-Gene files, RDML, qbase+ and Cq tables, matched to the plate you planned; raw curves kept |
| Analyze |  |  |
| Published methods | **Assembled** [NormqPCR](https://bioconductor.org/packages/release/bioc/html/NormqPCR.html) has geNorm, NormFinder and ΔΔCq; you supply the efficiencies, and its NRQ function is [marked experimental](https://bioconductor.org/packages/release/bioc/vignettes/NormqPCR/inst/doc/NormqPCR.pdf) | Stronger: **Built in** Efficiency per target from your dilution series, several reference genes, geNorm and NormFinder, a curve fit per well for Cq, error carried to the result |
| Your own methods | Stronger: **Anything** New methods reach R and Python first; any statistic, any plot | **Set methods** The published methods with the settings you choose; no custom code |
| Part of a pipeline | Stronger: **Native** Runs unattended next to sequencing, imaging or clinical data | **By file** RDML and Excel exports, file imports; no public API |
| Report |  |  |
| A report to hand over | Stronger: **Any format** [Quarto](https://quarto.org/docs/computations/r.html) runs your code inside the document and renders HTML, PDF or Word | **Report, notebook** A PDF or HTML report stamped with the Clarida release; the experiment in your Excel notebook; RDML 1.3 for the runs |
| Throughout |  |  |
| Who changed what | **Code edits** Git keeps each change to the script and its author; choices made while running it are not kept | **Decisions stored** Setting changes, automatic exclusions and accepted QC deviations, with who and when |
| One method for the whole lab | **Per copy** Each person’s copy can drift, unless the lab keeps one repository | Stronger: **Shared** Named settings sets that freeze once used; each experiment keeps its own copy; on a paid plan |
| The same numbers later | **If rebuilt** [renv](https://rstudio.github.io/renv/articles/renv.html) pins packages, not R or system libraries; [74% of shared R files](https://doi.org/10.1038/s41597-022-01143-6) failed to rerun | Stronger: **Frozen** A saved analysis never changes, even after an update |
| When its author leaves | **Needs a new owner** Packages too: [pcr](https://cran.r-project.org/package=pcr) left CRAN in 2025 when reported issues went unfixed | Stronger: **Maintained** Updated and tested by Clarida; nobody in your lab has to keep it running |

marks the stronger tool for that need.

### In short

- **Stronger in a script**: Your own methods · Part of a pipeline · A report to hand over
- **Stronger in Clarida**: Starting the next experiment · Mixes and volumes · Raw data in · Published methods · One method for the whole lab · The same numbers later · When its author leaves
- **A draw**: Plate layout · Filling the plate · Who changed what

## How do you check it against your script?

- **Start from a plate your script has already analyzed.** Import its run file, or the Cq table the script read.
- **Export RDML and read it back in R.** The [RDML package](https://cran.r-project.org/package=RDML) returns the Cq values and raw curves, well by well.
- **Compare per well, then per sample.** Cq, then efficiency, then the normalized quantity: a difference shows at the earliest step where the methods part.

## Can you keep your script beside it?

- **Plan in Clarida**: The [run layout](https://clarida.bio/qpcr/design#run-layout), the mixes and the bench notes, before your script takes over
- **Hand the run to your script**: [RDML 1.3](https://clarida.bio/qpcr/report#export-rdml) with the Cq values and raw curves, for the RDML package in R or [rdmlpython](https://pypi.org/project/rdmlpython/) in Python, which Clarida’s per-well efficiencies are tested against
- **Read the Excel export**: Each step in its own section with every intermediate value, ready for readxl or pandas ([Excel notebook](https://clarida.bio/qpcr/report#export-eln))
- **Bring your script’s Cq to Clarida**: A Cq table as Excel, CSV or text, its columns [mapped with you](https://clarida.bio/qpcr/import#import) before anything is saved
- **What does not carry over**: RDML holds no analysis settings; the Excel export names what it started from

## Frequently asked questions

### Isn’t a script more reproducible than any app?

For the code, yes: git and renv record exactly what ran. They don’t pin R itself or the system libraries, [as renv says](https://rstudio.github.io/renv/articles/renv.html), so rerunning years later can mean rebuilding an environment first. A [saved analysis](https://clarida.bio/methods/exploration-and-stored-analyses) in Clarida keeps its data, settings and results together and never changes, even after an update. A saved report carries the Clarida release and a hash of its input data. An analysis kept without one does not name the version yet.

### Can a script plan the plate as well?

Yes, and it can balance it. designit and Omixer spread treatment groups and other factors over plates, which matters in a large study where each one can act as a batch. Clarida does not randomize: it lays out the plate by grouping and fill rules for a multichannel pipette, then works out the mixes and guides you at the bench. Need both? Randomize in a script, then set Clarida’s [run layout](https://clarida.bio/methods/plate-layout-algorithm) to manual and place the wells as the script assigned them.

### Does a script you wrote need validating?

In regulated work, yes. OECD guidance for GLP counts statistical applications and data evaluation tools built for one facility as customized systems, which “bear the highest intrinsic risk”, and asks for a validation process and for the source code to be kept as long as the study records ([OECD No. 17](https://www.bfr.bund.de/cm/349/No17_Application-of-GLP-Principles-to-Computerised-Systems_2022.pdf)). Clarida needs validating as well, for the way your lab uses it, though nobody there maintains its code.

### What happens to your data if you stop using Clarida?

It stays yours, in open formats. RDML holds the raw curves, the Cq values and the plate, and the Excel workbook every result, with the settings it started from, so a script can take over where Clarida stopped.

### What does Clarida cost next to free packages?

Nothing, to start. The packages are free, and so is a first try. Clarida’s free plan: 4 experiments in your first month, then 2 a month. No card.

## Sources

Each source with the day we last checked it. Something out of date? [Tell us](https://clarida.bio/talk-to-us?topic=help).

- [Bustin et al. (2025), MIQE 2.0, Clinical Chemistry 71:634](https://doi.org/10.1093/clinchem/hvaf043) Report the analysis program and its version Checked October 26, 2026
- [Bioconductor, NormqPCR 1.58.0](https://bioconductor.org/packages/release/bioc/html/NormqPCR.html) Reference gene selection (geNorm, NormFinder) and normalization of qPCR data Checked October 29, 2026
- [Bioconductor, NormqPCR vignette](https://bioconductor.org/packages/release/bioc/vignettes/NormqPCR/inst/doc/NormqPCR.pdf) geNorm, NormFinder, ΔCq and ΔΔCq; its NRQ function takes your efficiencies and is marked experimental Checked October 26, 2026
- [CRAN, qpcR (release of June 10, 2025)](https://cran.r-project.org/package=qpcR) Sigmoidal fits of amplification curves, and model selection Checked October 29, 2026
- [CRAN, MCMC.qpcr 1.2.4 (November 8, 2025)](https://cran.r-project.org/package=MCMC.qpcr) Bayesian models for many genes at once, with controls as priors Checked October 29, 2026
- [Bioconductor, HTqPCR 1.66.0](https://bioconductor.org/packages/release/bioc/html/HTqPCR.html) High-throughput Ct values: loading, quality, normalization, testing Checked October 26, 2026
- [Bioconductor, ReadqPCR 1.58.0](https://bioconductor.org/packages/release/bioc/html/ReadqPCR.html) Reads raw RT-qPCR data of several platforms into R Checked October 26, 2026
- [CRAN, RDML 1.1 (July 23, 2025)](https://cran.r-project.org/package=RDML) Imports RDML files into R Checked October 26, 2026
- [PyPI, rdmlpython 1.7.2 (May 22, 2023)](https://pypi.org/project/rdmlpython/) A package to read and edit RDML files Checked October 29, 2026
- [CRAN, pcr (archived November 7, 2025)](https://cran.r-project.org/package=pcr) Archived "as issues were not corrected despite reminders" Checked October 29, 2026
- [Ahmed and Kim (2018), the pcr package, PeerJ 6:e4473](https://doi.org/10.7717/peerj.4473) Results checked against two published datasets when it appeared Checked October 26, 2026
- [Posit, renv](https://rstudio.github.io/renv/articles/renv.html) Records package versions in a lockfile; not R itself or the system libraries Checked October 26, 2026
- [Carver et al. (2022), research software in the United States, PeerJ Computer Science 8:e963](https://doi.org/10.7717/peerj-cs.963) Turnover among the most named difficulties; 29% include costs for maintaining software Checked October 26, 2026
- [Trisovic et al. (2022), research code quality and execution, Scientific Data 9:60](https://doi.org/10.1038/s41597-022-01143-6) 74% of over 9,000 shared R files failed on a fresh setup; 56% after automatic fixes Checked October 26, 2026
- [Gronenschild et al. (2012), software version and measurements, PLoS One 7:e38234](https://doi.org/10.1371/journal.pone.0038234) A new major release of an imaging package changed volumes by 8.8% on average Checked October 26, 2026
- [Bioconductor, release cycle](https://bioconductor.org/install/) A release every six months, each tied to one R version Checked October 26, 2026
- [OECD (2016, reissued 2022), GLP principles for computerized systems, Advisory Document No. 17](https://www.bfr.bund.de/cm/349/No17_Application-of-GLP-Principles-to-Computerised-Systems_2022.pdf) In-house analysis tools as customized systems, with a validation process; source code kept Checked October 26, 2026
- [CRAN, designit 0.5.1](https://cran.r-project.org/package=designit) Blocking and randomization of samples over batches and plates Checked October 26, 2026
- [Bioconductor, Omixer 1.22.0](https://bioconductor.org/packages/release/bioc/html/Omixer.html) Reproducible sample randomization across plates, with lab-friendly layouts Checked October 26, 2026
- [wellmap documentation](https://wellmap.readthedocs.io/en/latest/) Plate layouts in a text file that R and Python read Checked October 26, 2026
- [Opentrons, Python Protocol API](https://docs.opentrons.com/v2/) Python protocols that drive pipetting robots into well plates Checked October 26, 2026
- [Posit, Quarto with R](https://quarto.org/docs/computations/r.html) Runs R code inside a document, rendered as HTML, PDF or Word Checked October 26, 2026

Checked October 29, 2026

## Leave whenever you like, with the raw curves.

Analyze a plate, then export it as RDML 1.3 with its Cq values and raw curves, ready for the RDML package in R. 4 experiments in your first month, then 2 a month. No card.

[Start free](https://app.clarida.bio/login) · [Talk to a founder](https://clarida.bio/talk-to-us?topic=demo)
