Verified data-loading captures#

Every capture on this page came out of the running web application, at a fixed 1440×1000 viewport, and each crop is taken from the bounding rectangle of the control it shows rather than from hand-chosen pixels. The scenarios that reproduce them are in docs/_tooling/screenshots/scenarios.mjs.

The files opened here are the public Pancreas sample (scVelo’s E15.5 endocrinogenesis data) written in each format, and every link points at a server on the same computer, so the captures show what the app does with each format rather than with any one host.

Loading and session controls#

Close-up of File or folder: the full-width Open… button with its menu open, showing File… (AnnData, Seurat/SCE .rds/.rda, Loom, 10x .h5, .h5mu) and Folder… (Zarr, 10x outs, MatrixMarket), with the pointer on File…. — **Open…** under **File or folder**, with its menu open. **File…** opens a single-file chooser for every file format Cellucid reads; **Folder…** opens a folder chooser (a Zarr store, a Cell Ranger `outs/`, a MatrixMarket folder). A browser’s file dialog picks files or a folder, never both, which is why the button asks first.#

The Link or server bar holding the address of a cellucid serve server on this computer, with one line under it reading Server, the server's address and a compact Disconnect button. — The **Link or server** bar connected to a running `cellucid serve`: the bar holds its address (the port alone, `127.0.0.1:8765`, or the **Local URL** `http://127.0.0.1:8765` all reach it), and the line under it reads **Server** with **Disconnect**. The **Viewer URL** does not go here: its `?…` query makes the bar open it as a link to data. The Viewer URL belongs in the browser address bar. See Server Mode (CLI + Python).#

The Link or server bar with the GitHub path "theislab/cellucid-demo-custom-datasets/catalog" typed into it and the pointer on its Open button. — The **Link or server** bar holding a GitHub catalog folder path; **Open** reads its `datasets.json` (Host a Dataset on GitHub).#

Opening each format#

The whole Cellucid window after the Pancreas sample was opened as an .h5ad file with Open… › File…: the Session panel's Dataset box names the file, its cells and genes and "h5ad file · this computer", and the canvas shows its UMAP. — An `.h5ad` opened with **Open… › File…**. It is read in place, so only what a view needs is read; the Dataset box names the format and *this computer*.#

The Dataset box after a Zarr ZIP archive was opened from this computer: the dataset's name, its cells and genes, "Zarr v2 ZIP · this computer", its description, and Details open with its source and citation, the file, its format and size, how it is read, and what it holds. — A Zarr ZIP (`.zarr.zip`) opened from this computer, with **Details** open.#

The Dataset box after a Seurat object saved with saveRDS() was opened: its cells and genes, "R object (.rds) · this computer", the line "Read whole once (it cannot be read in parts), kept in this browser's storage for next time", and Details open with Read (whole, once, from start to end, nothing uploaded), Kept (the origin-private file system until it is cleared), Read in parts (a Cellucid-ready AnnData) and what it holds. — A Seurat object saved with `saveRDS()`. An R file has no random access, so it is read whole once into this browser’s storage and kept for next time, and the Dataset box says so on its own line (Open R objects and MatrixMarket folders).#

The opening panel over the viewer reading "Opening pancreas-objects.rda" and "pancreas-objects.rda holds 2 single-cell objects. Open which one?", with one button per Seurat object, "pancreas (Seurat, 3,696 cells)" and "pancreas_ductal (Seurat, 916 cells)", and Cancel. — An `.rda` holding two Seurat objects: the opening asks which one to open.#

The Dataset box after a Loom file was opened: the title stored in the file, its cells and genes, "Loom (.loom) · this computer", and Details open with the file, its size, how it is read, and what it holds: four obs fields and a UMAP. — A Loom file, read in place; its stored title names the dataset (Open Loom and 10x Cell Ranger files).#

The whole Cellucid window after a Cell Ranger outs/ folder was opened with Open… › Folder…: the Dataset box reads "10x Cell Ranger outs folder · this computer", and the canvas shows Cell Ranger's UMAP coloured by its clustering, gene_expression_clusters. — A Cell Ranger `outs/` folder opened with **Open… › Folder…**: the matrix is read in place, Cell Ranger’s UMAP is the embedding, and its clustering is a field.#

The Dataset box after a MatrixMarket folder with a cell table was opened with Open… › Folder…: the folder's name, its cells and genes, "MatrixMarket folder · this computer", the line saying where the converted data is kept, and Details open. — A MatrixMarket folder whose cell table holds a UMAP (Open R objects and MatrixMarket folders).#

The Data mapping group of a CITE-seq .h5mu with Genes open: a Modality select reading "rna · 28 features" above Expression, Gene names, Values and Highly variable genes, and Revert and Apply below. — A CITE-seq `.h5mu`: **Data mapping › Genes** shows a **Modality** select (Multimodal data: MuData and the Modality select).#

The opening itself#

The opening panel centred over the empty viewer while a linked Seurat .rds is downloaded: "Opening" and the file's address, the step "Downloading the whole file (9.4 MB): … does not serve parts of files", a progress bar part-way, how much has arrived with the speed and time left, the note saying the host serves the file only whole, how much is downloaded and that it is kept, and a Cancel button. — While nothing else is on screen, an opening is shown over the empty viewer: what is opened, the step under way, how much has arrived, and **Cancel**.#

The panel over the empty viewer after a linked file could not be opened: "The data could not be opened", one sentence saying why (the file was not found at its host, HTTP 404) and what to do, a folded Details, and the buttons Open without this link, focused, and Try again. — Data the page’s address names that cannot be opened stays as one panel over the empty viewer, with the reason, the error under **Details**, **Open without this link** and **Try again**. The rest of the app works.#

What a refusal looks like#

Each capture below is the notification Cellucid actually raised for that exact input, taken from the running application.

A Cellucid notification reading "The selected file is not a valid HDF5 file, so it cannot be an AnnData (.h5ad) file. Save it again with anndata…", with a dismiss button at its right. — A CSV renamed to `.h5ad`. The leading bytes are checked, not the extension.#

A Cellucid notification saying the Cellucid server could not be reached because nothing readable came back, that the connection may have failed or the address may not hold Cellucid data, and what to check. — A server address with nothing listening behind it.#

A Cellucid notification saying the linked data could not be opened because the GitHub folder holds no datasets.json, after a repository path rather than its catalog folder was opened. — A GitHub path pointing at a repository rather than at its catalog folder: the folder holds no `datasets.json`.#

The whole Cellucid window after a server address with nothing behind it was opened: the Pancreas sample is still drawn and still named in the Dataset box, and one error notification sits in the lower right. — None of these discard the dataset you already had. After a refusal the previous dataset is still open and still interactive.#

Server and notebook#

The server workflow is captured against real data in Server Mode (CLI + Python), with its Viewer URL, the complete --help output, and the common startup failures. The viewer a notebook embeds, and a real viewer.debug_connection() report, are in Jupyter Integration (Notebook Embedding).