# CyteType

> Annotate the clusters of single-cell RNA-seq data with cell types, Cell Ontology terms, cell states and reviewer confidence, from Python (Scanpy and AnnData, package `cytetype`) or R (Seurat, package `CyteTypeR`). This documentation is published from the master branch of the CyteType repository and covers both clients, the report, the dashboard, Chat, the MCP connector and target analysis.

- [Home](./): what CyteType does, a first Python and R example, and where to go next

## Getting started

- [Installation](getting-started/installation/): install the Python package or the CyteTypeR R package, check the installed version and keep it up to date
- [Sign in and API keys](getting-started/sign-in/): create an account, sign in from Python or R, use API keys on remote machines and in CI, and manage the saved key
- [Quickstart (Python)](getting-started/quickstart/): an executed notebook that preprocesses the PBMC 3k dataset with Scanpy, runs CyteType, and explores the labels, confidence and evidence
- [Quickstart (R)](getting-started/quickstart-r/): the same workflow in R: preprocess the PBMC 3k dataset with Seurat, run CyteTypeR, and explore the labels, confidence, evidence and report
- [How CyteType works](getting-started/how-cytetype-works/): what CyteType does with your clusters, the stages of a job, what you get back and what is uploaded

## Prepare and run

- [Prepare your data](guides/prepare-data/): get an AnnData or Seurat object ready, from normalized expression and cluster labels to marker genes, gene symbols, coordinates, raw counts and cell metadata
- [Large datasets](guides/large-datasets/): work from backed AnnData, rank marker genes in chunks, cap cells per cluster, and plan memory, disk space and uploads
- [Study context and metadata](guides/study-context/): write a study context, label runs with metadata tags, and choose the cell metadata columns that add evidence
- [Run, monitor and rerun jobs](guides/running-jobs/): what `run()` and `CyteTypeR()` do from upload to results, how to follow a job, and how to fetch results later and rerun jobs
- [Use your own LLM](guides/llm-configuration/): run the annotation agents on your own model provider, including AWS Bedrock, OpenAI-compatible endpoints and local Ollama models
- [Automation, HPC and CI](guides/automation/): run CyteType from scripts, HPC batch jobs and CI, with an API key from a secret store and saved results

## Results

- [Results in AnnData and Seurat](results/results-in-your-data/): what CyteType adds to AnnData and Seurat objects, what `get_results()` and `GetResults()` return, and how to save, plot and export the results
- [The CyteType report](results/report/): the interactive report, from following a running job to reading every evidence section, comparing with your own labels and exporting a CSV
- [Interpret and review results](results/interpreting-results/): read labels, reviewer confidence, heterogeneity flags and classifications, know the limits of the results, and review a run with a checklist

## Platform

- [Dashboard](platform/dashboard/): manage your reports, API keys and account, in your personal or your organization's workspace
- [Visibility and sharing](platform/sharing/): private, organization and public reports, email sharing, and what each level exposes
- [Chat](platform/chat/): ask questions about completed runs and get answers, tables and figures from the stored results and uploaded data
- [Target analysis](platform/target-analysis/): rank cell-type-resolved target hypotheses for a completed job, and read the results and their limits
- [Connect AI assistants (MCP)](platform/mcp/): connect Claude, Claude Code, Cursor or another MCP client to explore and compare your annotation jobs
- [Usage limits and quotas](platform/limits/): limits for personal accounts and organizations, the data, job, Chat and MCP limits, and the errors you get when you reach them

## Reference

- [Python API](api/index.html): exact signatures and docstrings of the `cytetype` package
- [CyteType class](api/cytetype.html): prepare an AnnData object, run a job, read the results and delete the upload files
- [Helpers](api/helpers.html): `marker_dotplot`, `rank_genes_groups_backed` and `subsample_by_group`
- [LLM configuration](api/llm-config.html): the fields of each dict in `llm_configs`
- [Exceptions](api/exceptions.html): the exception classes of the Python client and their error codes
- [R function reference](reference/r-api/): the exported functions of the CyteTypeR package, grouped into preparing and running jobs, results, sign-in and errors, with their usage, arguments, return values and examples, generated from the help pages of CyteTypeR 0.6.0
- [Command-line reference](reference/cli/): the `cytetype` (Python) and `cytetyper` (R) commands with every option, environment variable, output and exit code
- [Errors and error codes](reference/errors/): the exceptions of the Python client and the conditions of CyteTypeR, the server error codes behind them, and what to do about each

## Support

- [Troubleshooting](support/troubleshooting/): fixes for common problems with signing in, preparing data, uploading files, running jobs and installing the clients
- [FAQ](support/faq/): short answers about data requirements, analysis, privacy, access, pricing and support
- [Security and privacy](support/security-and-privacy/): what the clients upload, how CyteType protects and uses your data, who can see your reports, and how deletion and retention work
- [Cite CyteType and report versions](support/citing/): how to cite CyteType, what to record so that others can reproduce your annotations, and the licenses of the clients and the service
