CyteType

CyteType annotates the clusters of single-cell RNA-seq data. You provide the clusters and marker genes you have computed, from Python with Scanpy and AnnData or from R with Seurat, and CyteType returns for each cluster a cell type label, a Cell Ontology term, a cell state and a confidence rating from an independent review step. The annotation runs on CyteType’s servers, and an interactive report shows the evidence behind every label.

A first run¶

Install the client, sign in once, then annotate a clustered dataset. Using CyteType requires an API key: cytetype setup or SetupCyteTypeR() signs you in with your browser and saves your key on this computer.

pip install cytetype
cytetype setup

import scanpy as sc
from cytetype import CyteType

# adata holds log-normalized expression and Leiden clusters in adata.obs["leiden"].
sc.tl.rank_genes_groups(adata, groupby="leiden", key_added="rank_genes_leiden")

annotator = CyteType(adata, group_key="leiden", rank_key="rank_genes_leiden")
adata = annotator.run(
    study_context="Peripheral blood mononuclear cells from a healthy human donor."
)

install.packages("devtools")
devtools::install_github("NygenAnalytics/CyteTypeR")
CyteTypeR::SetupCyteTypeR()

library(Seurat)
library(CyteTypeR)

# obj holds normalized data in its RNA assay, a UMAP and clusters from
# FindClusters().
markers <- FindAllMarkers(obj, only.pos = TRUE)
prepped <- PrepareCyteTypeR(obj, markers, group_key = "seurat_clusters")
obj <- CyteTypeR(
  obj,
  prepped,
  study_context = "Peripheral blood mononuclear cells from a healthy human donor."
)

The link to the report is printed as soon as the job starts. The Python quickstart and the R quickstart walk through a complete example.

Where to go next¶

• Getting started: install the client, sign in, and read how CyteType works.

• Prepare and run: prepare your data, work with large datasets, write a study context, run, monitor and rerun jobs, use your own LLM, and run CyteType from scripts, HPC and CI.

• Results: find the results in AnnData and Seurat, read the CyteType report, and interpret and review the results.

• Platform: manage your jobs in the dashboard, control visibility and sharing, ask questions about your results in Chat, run a target analysis, connect AI assistants through MCP, and check the usage limits.

• Reference: the Python API, the R functions, the command line and the error codes.

• Support: troubleshooting, the FAQ, security and privacy and how to cite CyteType.

Source code and issues¶

The Python client is developed in NygenAnalytics/CyteType and the R package in NygenAnalytics/CyteTypeR. Report problems in the CyteType issues or the CyteTypeR issues. The documentation index for LLMs lists every page in plain text for assistants that read this site.