Dashboards
Compose, coordinate and share multi-panel dashboards from a single portable JSON spec — a no-code builder that is also a scientific analysis environment.
What the Dashboards Layer Is
The canvasxpress-dashboards package turns individual charts into coordinated dashboards described by one portable JSON spec. It reuses the same broadcast engine, so a selection or filter in one panel updates the others with no extra code. You can build a dashboard in the browser with a no-code builder, bind its panels to live data through connectors, and persist, publish and share the result — and because the dashboard is just a spec plus data, it stays reproducible.
Try the live builder right in your browser, or grab the package on GitHub.
No-Code Builder
The builder lets you drag, resize and edit each panel with CanvasXpress’s own customizer — the same interactive UI that drives every chart. A wizard can lay out a whole dashboard from one table, so you start from a working board rather than an empty canvas. Every panel is a standard CanvasXpress figure, so anything you can author in a single chart is available inside a dashboard panel.
Coordinated Panels
Panels are linked through the broadcast engine: a selection, hover or filter in one panel is broadcast to the others, so the whole board responds together with no custom wiring. Filter and marking schemes let a dashboard behave like a coordinated analysis surface rather than a set of independent charts.
Live Data Sources
Panels can be bound to authenticated databases and Google Sheets through the connectors package, which keeps credentials on the server rather than exposing them to the browser. Dashboards can be persisted, published and shared with pluggable storage — local files, S3, Postgres or Google Drive.
A Scientific Analysis Environment
Beyond BI-style composition, the dashboards layer is built as a scientific analysis environment — a set of capabilities no general-purpose dashboard tool offers:
Reproducibility & repro bundles
Every dashboard records its provenance and exports a self-contained
bundle — data, spec and the dataToR/dataToPython code to regenerate it —
so a figure is something you can defend and rebuild, not a static picture.
Domain-native science panels
Kaplan–Meier, oncoprint/MAF-matrix, genome/Manhattan and volcano panels are auto-detected from the data and wired by the wizard — plus enrichment and VCF ingestion, dependency-free.
Publication figure composer
Assemble multi-panel figures and export them as vector SVG/PDF at a chosen DPI with publication theming — the daily “lay it out in Illustrator” step, done in the browser.
Statistical inference
Pairwise significance brackets with p-values and stars, native multiple-testing correction (BH / Bonferroni / BY / Holm / Hochberg), promoted volcano thresholds and effect-size + CI tiles. Read the guide →
Analysis controls
Transform and normalization controls (log, z-score, normalize, R-verified) and an FDR slider that re-filters every significance panel live — interactive analysis, not just interactive charts.
Packages & Get Started
The dashboards layer ships as open-source packages for both ecosystems — canvasXpressDashboards for R and canvasxpress-dashboards for Python — alongside the in-browser builder. The portable JSON spec is the common format: author a dashboard in the builder, save the spec, and recreate or version it from code. Start from the live builder, or read the source on GitHub.
