Web App Guide#
This guide documents the Cellucid web app UI end-to-end: how to load data, navigate, filter, analyze, export figures, share sessions, and troubleshoot.
If you are looking for Python-side APIs (prepare, serve, show_anndata, hooks), start with Python Package Guide instead.
Where to start (pick one)#
First time using Cellucid → Landing / Orientation Pages (Start Here) then Data Loading in the Web App (All Paths)
“I just need to load my dataset” → Data Loading in the Web App (All Paths)
“I need to understand filtering + selection” → Filtering (Visibility, Outliers, and Filter Stacks) and Highlighting and Selection (Groups, Pages, Tools)
“I want RNA velocity / vector overlays” → Vector Field / Velocity Overlay (GPU Particle Overlay)
“I want to undo, go back to an earlier step, or write up what I did” → State and History
“I want Claude or ChatGPT to work in Cellucid with me” → Agents
“I want community annotation workflows” → Community Annotation (Voting + Consensus; GitHub Sync)
“Something is broken” → Troubleshooting (Web App): Start Here
Chapters (what each folder contains)#
Getting started#
What Cellucid is, system requirements, a 60-second tour, and how to choose a workflow.
All supported loading paths (file picker, links to GitHub, GEO, Zenodo and other hosts, server mode, Python and R viewers), the formats Cellucid reads (AnnData, MuData, Loom, 10x, MatrixMarket, R objects), dataset identity, data mapping, and troubleshooting.
Core UI workflows#
Navigation, selection primitives, and the “mental model” of interacting with points in the viewer.
How Cellucid treats obs fields, categorical vs continuous coloring, legends, and common field pitfalls.
Filtering mental model, UI controls, and edge cases (including “why did my selection disappear?”).
Selecting points, highlighting, synchronization rules, and how to reason about “active vs visible vs selected”.
Analysis panel workflows (e.g. marker genes / differential expression), interpretation guidance, and troubleshooting.
Claude, ChatGPT and local agent apps working in Cellucid: the chat bar, the connector, your permissions, Stop and Undo turn, and what agents are sent of your data.
Specialized features#
Enable overlays, choose fields per dimension, tune parameters, and troubleshoot missing/incorrect velocity visuals.
How multi-user annotation voting works, author setup, annotator workflows, and UI reference.
Export UI walkthrough, formats, quality knobs, metadata/provenance, and edge cases for publication-ready figures.
The session file (your history in one file), share links, collaboration patterns, and “what makes a session compatible”.
Every step you take, kept as a history: undo and redo, branches, trackers, saved states in this browser, a guided tour, and pseudocode, a methods paragraph or a printed report of what you did.
Performance, safety, and maintenance#
Performance mental model, best practices for large datasets, and how to diagnose GPU/rendering bottlenecks.
Accessibility guidance plus a practical privacy/security model for “browser ↔ server ↔ data”.
Architecture notes, state/event model, debugging playbooks, and extension points for contributors.
Symptom → diagnosis → fix across installation, data loading, rendering, selection, analysis, and export.