Plöttr — A browser-only data-analysis toolbox

Static web app for common plot-and-test workflows. No server, no tracking, no data leaving the browser.

Live: evompmi.github.io/plottr · Source: github.com/evompmi/plottr

Originally built for the "Evolution of plant-microbes interactions" team members at Toulouse Plant Sciences (University of Toulouse, France). Released publicly for anyone with similar needs.

Project status

Plöttr is stable and feature-complete as of v1.9.0. It is maintained on an as-needed basis — issues and pull requests are welcome, but no major new work is planned. Because it ships as static files with no runtime dependencies and no backend, it is designed to keep running indefinitely, and every release is archived for citation via Zenodo. Fork freely under the MIT license.

Design goals

• Zero infrastructure. Static HTML/CSS/JS. Runs from GitHub Pages or any static-file server (python3 -m http.server, npx serve, nginx, …) — no backend.
• Privacy by construction. All parsing and computation happens in-browser. Works offline once loaded.
• Paste-and-plot. Each tool auto-detects separators (, / \t) and decimal convention (, / .), then walks through column roles → filter → plot → export.
• Publication-ready output. SVG (named <g> groups for Inkscape/Illustrator), PNG (2×), CSV of processed data.
• Honest statistics. Welch's t / Welch ANOVA + Games-Howell are the default picks (Rasch et al. 2011; Zimmerman 2004 — pre-screening with Shapiro-Wilk to choose between parametric and non-parametric is a known anti-pattern). Shapiro-Wilk and Levene are still computed and shown as diagnostics in the decision trace; the trace explains why Welch is the default and how to override per-test.

Tools

Tool	Purpose
RLU timecourse	Luminescence time-course with mean ± SD, per-replicate integrals, inline stats, and optional aequorin Ca²⁺ calibration (Allen & Blinks / Hill / Generalised).
Group Plot	Box / violin / raincloud / bar with Welch t / Welch ANOVA + Games-Howell as the default pick (override to Student / one-way ANOVA / Mann–Whitney / Kruskal–Wallis from the stats panel).
Line Plot	Mean ± SEM / SD / 95 % CI per group across a shared x, with per-x significance markers.
Scatter Plot	XY with colour / size / shape mapping, reference lines, and optional linear regression overlay.
Heatmap	Matrix heatmap with row / column clustering (hierarchical, k-means), dendrograms, and zoomed detail view.
Venn Diagram	2–3 set area-proportional Venn with click-to-extract region members.
UpSet Plot	4+ set intersection plot — top bar chart, matrix of participation dots, per-set totals. Click a column to inspect its items.
Volcano Plot	log₂FC vs −log₁₀(p) for −omics hits with three-class significance colouring, top-N labels with collision avoidance, click-to-label, search-by-name, optional colour / size aesthetic mappings, classified CSV + ggplot2 R-script export. Auto-detects DESeq2 / limma / edgeR column conventions.
Proportions	Counts table → % per group with Wilson 95 % CIs, stacked or mosaic bars, Fisher's exact / χ² test (auto-picked by Cochran's rule), Holm-corrected pairwise Fisher tests, goodness of fit to ratios such as 3:1 or 9:3:3:1.
qPCR	Raw Ct → relative expression: 2^−ΔΔCt or efficiency-corrected (Pfaffl) ratios, several reference genes by geometric mean, efficiency from a standard curve, technical-replicate QC, geNorm M. Statistics run on ΔCt (never on fold changes); fold change plotted on a log₂ axis; ΔCt can be sent to Group Plot.
Power Analysis	A-priori and post-hoc power for t (indep / paired / one-sample), one-way ANOVA, χ², correlation.
Calculator	Molarity, dilution (C₁V₁ = C₂V₂), ligation ratio, batch prep sheets.

Each tool has an in-app How to panel.

Screenshots

Figures as exported (PNG):

Statistical validation

All numerics (tools/_core/stats/*.ts) are cross-validated against two independent references:

• R 4.5 on real built-in datasets (iris, PlantGrowth, ToothGrowth, mtcars, …). Current run: 618 comparisons, 0 failures, including 4 rows past R's precision floor, where R's ptukey cancels to ~2.2 × 10⁻¹⁵ while Plöttr's ptukey_upper continues the true tail (cross-checked against SciPy and Monte Carlo). Reproducible via npm run benchmark. Results render as a public page at benchmark.html — failing rows would be shown in red, not hidden.
• SciPy 1.17 on synthetic targeted grids over the (df, λ) regimes the R bank only touches indirectly: nctcdf at deep δ, ncf_sf / ncchi2cdf at large λ across the noncentral normal-approx threshold, qtukey at extreme corners. 1,083 comparisons, 0 failures, with 92 deep-tail / underflow rows (both values < 10⁻¹³, informational) and 1 pathological annotation (qtukey at df = 1, documented as outside the design envelope in tools/_core/stats/posthoc.ts). Reproducible via npm run benchmark:scipy.

On top of that, a deterministic test suite of 2,100+ tests (Vitest 4, CI-gated on every commit alongside ESLint, Prettier, tsc --noEmit, and npm run build) spanning unit, integration, render-smoke (real React 18 + happy-dom), and per-tool property suites (fast-check, with automatic shrinking on failure). A Playwright e2e suite (npm run e2e) covers paste → configure → plot golden paths per tool. Mutation testing is wired up via Stryker (npm run mutation) for on-demand meta-tests of whether the suite catches bugs, not just whether tests pass.

Covers: Shapiro–Wilk, Brown–Forsythe Levene, Student / Welch t, Mann–Whitney U, one-way ANOVA, Welch ANOVA, Kruskal–Wallis, Pearson r / Spearman ρ / Kendall τ-b correlations, Tukey HSD (studentised range), Games–Howell, Dunn + Benjamini–Hochberg, Cohen's d (with 95 % CI), Hedges' g, η², ε², compact letter display.

Scope and limitations

Numerics are trustworthy, but the design covers a slice of real workflows.

Fits best: one-way group comparisons with a defensible test pick; quick exploratory plotting from pasted CSV/TSV; privacy-sensitive data; publication-ready SVG; a-priori power analysis; reproducibility via the one-click R script export (every statistics tool, plus Heatmap, Volcano and Power Analysis, emits a runnable R script that embeds the data inline and reproduces the exact analysis); niche aequorin Ca²⁺ calibration not first-class anywhere else.

You will outgrow it for: repeated-measures / mixed models, factorial designs (no two-way ANOVA, no interactions, no ANCOVA), Dunnett's test, multiple / logistic / non-linear regression, survival / ROC / time-to-event, large datasets (browser-only, millions of rows won't work), headless batch processing. The auto-pick defaults to Welch's t / Welch ANOVA so it doesn't gate on Shapiro-Wilk; per-group SW at α = 0.05 still inflates the family-wise diagnostic flag rate at large k (~23 % at k = 5 even with everything genuinely normal), but that flag is now a non-parametric suggestion in the trace, not a routing decision. Tunable via alphaNormality.

Use as a supplement to R / Prism / SPSS, not a replacement.

Installation

For casual use, visit evompmi.github.io/plottr.

To run offline or modify the source:

git clone https://github.com/evompmi/plottr.git
cd plottr
python3 -m http.server     # or `npx serve` — any static-file server works
# Then open http://localhost:8000 in any modern browser.

(Direct file:// no longer works as of v1.3.0: the SPA shell loads tools/_app/index.js as a JavaScript module, and browsers block module loading from file:// origins.)

Development

Node.js ≥ 20 for the tooling (not for running the app):

npm install
npm run build       # compile tools/*.tsx → tools/*.js
npm run watch       # recompile on save
npm test            # full deterministic suite (Vitest 4)
npm run typecheck   # tsc --noEmit
npm run lint        # ESLint
npm run benchmark   # R + SciPy cross-validation (R 4.5, SciPy 1.17)
npm run e2e         # Playwright e2e suite
npm run mutation    # Stryker mutation testing (on demand)

Architecture, shared-code constraints, and conventions for adding tools are in CLAUDE.md. Versioned changes are tracked in CHANGELOG.md; per-release long-form notes live under docs/release-notes/.

Technical stack

React 18 (vendored under vendor/, no CDN) + esbuild (TSX → JS, ~5 ms incremental) + custom SVG rendering (no charting library) + in-house stats cross-validated against R 4.5 + SciPy 1.17. Plot tools share a typed TypeScript scaffold under tools/_shell/ (step navigator, upload + paste handlers, style-prefs persistence, stats dispatch); each tool's App ships as a lazy chunk, so the SPA only downloads the route you visit. Tests via Vitest 4 with a thin compat shim that preserves the project's house vocabulary. Hosted on GitHub Pages as a single-page app (hash-routed; one esbuild entry plus per-tool code-split chunks).

No runtime dependencies from external CDNs — vendor/ ships React + ReactDOM so a cloned copy works without network access.

Citing

If you use Plöttr in published research, please cite the repository. Suggested citation:

Plöttr — a browser-only data-analysis toolbox for wet-lab scientists. https://github.com/evompmi/plottr. M. Mbengue, Laboratoire de Recherche en Sciences Végétales (LRSV), Université de Toulouse, CNRS, UPS, Toulouse INP, Castanet-Tolosan, France.

License

Plöttr is released under the MIT License. Vendored copies of React + ReactDOM (vendor/) keep their upstream MIT license — see vendor/LICENSE-react.txt. Continued-fraction primitives (betacf, gammainc, gammainc_upper in tools/_core/stats/dist.ts) are ported from the public-domain Cephes Mathematical Library. All algorithmic references and third-party attributions are consolidated in THIRD_PARTY.md.

AI Usage

Plöttr was built collaboratively with Anthropic's Claude (Anthropic Claude Opus 4.8 [Large language model]. https://claude.ai) under researcher direction. The bulk of the JavaScript, tests, and documentation is Claude-written; the researcher specifies needs, reviews proposals, picks scope, sets methodology, and approves every commit before it lands. Architectural conventions are encoded in folder-level CLAUDE.md files so the workflow is reproducible.

What is independently verified. Every statistical function is cross-validated against R 4.5.3 and SciPy 1.17.1 on each npm run benchmark. Mutation testing (npm run mutation, Stryker) probes whether the test suite catches regressions. These checks exist precisely because the implementation is AI-generated.

What is not independently verified and rests on researcher review alone: UI / UX choices, accessibility labels, the in-app How-to panels and the chart aesthetics. Bug reports there are especially welcome — the "Send feedback" button in every tool opens an in-app dialog with the address and a prefilled message you can open in your mail app or copy into webmail.

References

Plöttr's statistical, numerical, and visualisation methods are independent implementations of published algorithms and methodology. Listed below is every research reference behind the codebase — swept from the source comments and verified against the publishers (Crossref, PubMed, journal pages). Code provenance and licensing for vendored and ported code (React, the Cephes ports, the colour palettes) is detailed separately in THIRD_PARTY.md.

A few entries are books or pre-DOI papers with no DOI; those carry an archival or publisher link where one exists, or none at all.

Statistical tests and post-hoc procedures

• Welch, B. L. (1947). The generalization of "Student's" problem when several different population variances are involved. Biometrika 34(1–2), 28–35. https://doi.org/10.1093/biomet/34.1-2.28
• Satterthwaite, F. E. (1946). An approximate distribution of estimates of variance components. Biometrics Bulletin 2(6), 110–114. https://doi.org/10.2307/3002019
• Levene, H. (1960). Robust tests for equality of variances. In I. Olkin et al. (Eds.), Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling (pp. 278–292). Stanford University Press. https://archive.org/details/contributionstop0000unse_d2c5
• Brown, M. B., & Forsythe, A. B. (1974). Robust tests for the equality of variances. Journal of the American Statistical Association 69(346), 364–367. https://doi.org/10.1080/01621459.1974.10482955
• Royston, P. (1992). Approximating the Shapiro-Wilk W-test for non-normality. Statistics and Computing 2(3), 117–119. https://doi.org/10.1007/BF01891203
• Royston, P. (1995). Remark AS R94: A remark on Algorithm AS 181: The W-test for normality. Journal of the Royal Statistical Society, Series C (Applied Statistics) 44(4), 547–551. https://doi.org/10.2307/2986146
• Blom, G. (1958). Statistical Estimates and Transformed Beta-Variables. Almqvist & Wiksell / John Wiley & Sons. (Monograph; no DOI.)
• Lehmann, E. L. (1975). Nonparametrics: Statistical Methods Based on Ranks. Holden-Day. https://archive.org/details/nonparametricsst0000lehm
• Siegel, S., & Castellan, N. J. (1988). Nonparametric Statistics for the Behavioral Sciences (2nd ed.). McGraw-Hill. (Textbook; no DOI.)
• Dunn, O. J. (1964). Multiple comparisons using rank sums. Technometrics 6(3), 241–252. https://doi.org/10.1080/00401706.1964.10490181

Effect sizes

• Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum Associates. (2013 Routledge reissue: https://doi.org/10.4324/9780203771587)
• Hedges, L. V. (1981). Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics 6(2), 107–128. https://doi.org/10.3102/10769986006002107
• Cumming, G., & Finch, S. (2001). A primer on the understanding, use, and calculation of confidence intervals that are based on central and noncentral distributions. Educational and Psychological Measurement 61(4), 532–574. https://doi.org/10.1177/0013164401614002
• Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Frontiers in Psychology 4, Article 863. https://doi.org/10.3389/fpsyg.2013.00863
• Tomczak, M., & Tomczak, E. (2014). The need to report effect size estimates revisited. An overview of some recommended measures of effect size. Trends in Sport Sciences 21(1), 19–25. (No DOI; published by AWF Poznań.)

Test-selection methodology

• Hoenig, J. M., & Heisey, D. M. (2001). The abuse of power: the pervasive fallacy of power calculations for data analysis. The American Statistician 55(1), 19–24. https://doi.org/10.1198/000313001300339897
• Zimmerman, D. W. (2004). A note on preliminary tests of equality of variances. British Journal of Mathematical and Statistical Psychology 57(1), 173–181. https://doi.org/10.1348/000711004849222
• Schucany, W. R., & Ng, H. K. T. (2006). Preliminary goodness-of-fit tests for normality do not validate the one-sample Student t. Communications in Statistics – Theory and Methods 35(12), 2275–2286. https://doi.org/10.1080/03610920600853308
• Rasch, D., Kubinger, K. D., & Moder, K. (2011). The two-sample t test: pre-testing its assumptions does not pay off. Statistical Papers 52(1), 219–231. https://doi.org/10.1007/s00362-009-0224-x
• Delacre, M., Leys, C., Mora, Y. L., & Lakens, D. (2019). Taking parametric assumptions seriously: arguments for the use of Welch's F-test instead of the classical F-test in one-way ANOVA. International Review of Social Psychology 32(1), Article 13. https://doi.org/10.5334/irsp.198

Numerical methods and special functions

• Wilson, E. B., & Hilferty, M. M. (1931). The distribution of chi-square. Proceedings of the National Academy of Sciences 17(12), 684–688. https://doi.org/10.1073/pnas.17.12.684
• Cornish, E. A., & Fisher, R. A. (1938). Moments and cumulants in the specification of distributions. Revue de l'Institut International de Statistique 5(4), 307–320. https://doi.org/10.2307/1400905
• Abramowitz, M., & Stegun, I. A. (Eds.) (1964). Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. National Bureau of Standards, Applied Mathematics Series 55. https://www.nist.gov/mathematics-statistics/handbook-mathematical-functions-abramowitz-and-stegun
• Lanczos, C. (1964). A precision approximation of the gamma function. Journal of the Society for Industrial and Applied Mathematics, Series B: Numerical Analysis 1(1), 86–96. https://doi.org/10.1137/0701008
• Welford, B. P. (1962). Note on a method for calculating corrected sums of squares and products. Technometrics 4(3), 419–420. https://doi.org/10.1080/00401706.1962.10490022
• Park, S. K., & Miller, K. W. (1988). Random number generators: good ones are hard to find. Communications of the ACM 31(10), 1192–1201. https://doi.org/10.1145/63039.63042
• Godfrey, P. (2001). A note on the computation of the convergent Lanczos complex Gamma approximation. Unpublished technical note. https://www.numericana.com/answer/info/godfrey.htm
• Acklam, P. J. An algorithm for computing the inverse normal cumulative distribution function. Undated web note; the original site (home.online.no/~pjacklam) is offline and archived via the Internet Archive Wayback Machine.

Clustering

• Lance, G. N., & Williams, W. T. (1967). A general theory of classificatory sorting strategies: 1. Hierarchical systems. The Computer Journal 9(4), 373–380. https://doi.org/10.1093/comjnl/9.4.373
• Lloyd, S. P. (1982). Least squares quantization in PCM. IEEE Transactions on Information Theory 28(2), 129–137. https://doi.org/10.1109/TIT.1982.1056489
• Arthur, D., & Vassilvitskii, S. (2007). k-means++: the advantages of careful seeding. Proceedings of the 18th Annual ACM-SIAM Symposium on Discrete Algorithms (SODA '07), 1027–1035. https://dl.acm.org/doi/10.5555/1283383.1283494

Set statistics

• Wang, M., Zhao, Y., & Zhang, B. (2015). Efficient test and visualization of multi-set intersections. Scientific Reports 5, Article 16923. https://doi.org/10.1038/srep16923

Visualisation

• Heckbert, P. S. (1990). Nice numbers for graph labels. In A. S. Glassner (Ed.), Graphics Gems (pp. 61–63). Academic Press. https://publications.ri.cmu.edu/nice-numbers-for-graph-labels/
• Silverman, B. W. (1986). Density Estimation for Statistics and Data Analysis. Chapman & Hall. (2018 CRC Press reprint: https://doi.org/10.1201/9781315140919)
• Wong, B. (2011). Points of view: color blindness. Nature Methods 8(6), 441. https://doi.org/10.1038/nmeth.1618
• Nuñez, J. R., Anderton, C. R., & Renslow, R. S. (2018). Optimizing colormaps with consideration for color vision deficiency to enable accurate interpretation of scientific data. PLOS ONE 13(7), e0199239. https://doi.org/10.1371/journal.pone.0199239

Aequorin calcium calibration

• Allen, D. G., & Blinks, J. R. (1978). Calcium transients in aequorin-injected frog cardiac muscle. Nature 273(5663), 509–513. https://doi.org/10.1038/273509a0
• Knight, M. R., Campbell, A. K., Smith, S. M., & Trewavas, A. J. (1991). Transgenic plant aequorin reports the effects of touch and cold-shock and elicitors on cytoplasmic calcium. Nature 352(6335), 524–526. https://doi.org/10.1038/352524a0
• Plieth, C. (2006). Aequorin as a reporter gene. Methods in Molecular Biology 323, 307–327. https://doi.org/10.1385/1-59745-003-0:307

Datasets

• Anderson, E. (1935). The irises of the Gaspé Peninsula. Bulletin of the American Iris Society 59, 2–5. (No DOI.)
• Fisher, R. A. (1936). The use of multiple measurements in taxonomic problems. Annals of Eugenics 7(2), 179–188. https://doi.org/10.1111/j.1469-1809.1936.tb02137.x