Skip to content

Visualizations

Celldega ships interactive visualizations, grouped into three kinds:

Every visualization accepts an optional name= (or registry_key=). Creating another widget of the same kind with the same key closes and replaces the old one, which is useful when rerunning notebook cells that create GPU-backed widgets. Use distinct keys when widgets should coexist. Widgets without a key coexist by default; Clustergram retains its matrix-name identity and Enrich retains its existing "default" identity.

Spatial

Render directly in geographic (x/y, and sometimes z) space over a tissue section, using deck.gl for GPU-accelerated rendering of large point/polygon datasets.

  • Landscape — the main spatial view: cells, transcripts, images, and neighborhoods over a tissue section.
  • Yearbook — a grid of per-cell spatial "portraits" cropped from the same underlying LandscapeFiles.
  • CellCloud — a 3D orbit-camera view of cell centroids, for thick tissue or multi-slice alignments.
  • NeighborhoodCloud — a 3D orbit-camera view of precomputed tissue neighborhoods.

Data

Render a dataset's values directly, independent of spatial position.

  • Clustergram — a hierarchically clustered heatmap (dendrograms, reorderable rows/columns) over a matrix (e.g. genes by cells or genes by clusters).
  • Composition — a Clustergram variant comparing category composition (e.g. cell-type proportions) across groups.

Info

Summarize or look up information about a gene list rather than rendering a dataset's cells or matrix directly.

  • Enrich — gene set enrichment analysis against public libraries (via the Enrichr API).

Note

Screenshots and example videos for each visualization are coming soon. For now, each page below documents what the visualization shows and how to create it; see the Example Notebooks and Gallery for runnable demos.