⚠️ CyteType is under active development and breaking changes may be introduced. Please work with the latest version to ensure compatibility and access to new features.
CyteType is a Python package for deep characterization of cell clusters from single-cell RNA-seq data. This package interfaces with Anndata objects to call CyteType API.
- Example Report
- Quick Start
- Installation
- Usage
- Configuration Options
- Annotation Process
- Development
- License
View a sample annotation report: CyteType Report
import anndata
import scanpy as sc
import cytetype
# Load and preprocess your data
adata = anndata.read_h5ad("path/to/your/data.h5ad")
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)
sc.pp.pca(adata)
sc.pp.neighbors(adata)
sc.tl.leiden(adata, key_added = "clusters")
sc.tl.rank_genes_groups(adata, groupby='clusters', method='t-test')
# Initialize CyteType (performs data preparation)
annotator = cytetype.CyteType(adata, group_key='clusters')
# Run annotation
adata = annotator.run(
study_context="Human brain tissue from Alzheimer's disease patients"
)
# View results
print(adata.obs.cytetype_annotation_clusters)
print(adata.obs.cytetype_cellOntologyTerm_clusters)pip install cytetypeYour AnnData object must have:
- Log-normalized expression data in
adata.X - Cluster labels in
adata.obs - Differential expression results from
sc.tl.rank_genes_groups
import scanpy as sc
# Standard preprocessing
sc.pp.normalize_total(adata, target_sum=1e4)
sc.pp.log1p(adata)
# Clustering
sc.pp.pca(adata)
sc.pp.neighbors(adata)
sc.tl.leiden(adata, key_added='clusters')
# Differential expression (required)
sc.tl.rank_genes_groups(adata, groupby='clusters', method='t-test')from cytetype import CyteType
# Initialize (data preparation happens here)
annotator = CyteType(adata, group_key='clusters')
# Run annotation
adata = annotator.run(
study_context="Adult human brain tissue samples from healthy controls and Alzheimer's disease patients, analyzed using 10X Genomics single-cell RNA-seq. Samples include cortical and hippocampal regions."
)
# Or with custom metadata for tracking
adata = annotator.run(
study_context="Adult human brain tissue samples from healthy controls and Alzheimer's disease patients, analyzed using 10X Genomics single-cell RNA-seq. Samples include cortical and hippocampal regions.",
metadata={
'experiment_name': 'Brain_AD_Study',
'run_label': 'initial_analysis'
}
)
# Results are stored in:
# - adata.obs.cytetype_annotation_clusters (cell type annotations)
# - adata.obs.cytetype_cellOntologyTerm_clusters (cell ontology terms)
# - adata.uns['cytetype_results'] (full API response)annotator = CyteType(
adata,
group_key='leiden', # Required: cluster column name
rank_key='rank_genes_groups', # DE results key
gene_symbols_column='gene_symbols', # Gene symbols column
n_top_genes=50, # Top marker genes per cluster
aggregate_metadata=True, # Aggregate metadata
min_percentage=10, # Min percentage for cluster context
pcent_batch_size=2000, # Batch size for calculations
coordinates_key='X_umap', # Coordinates key for visualization
max_cells_per_group=1000, # Max cells per group for visualization
)The run method accepts several configuration parameters to control the annotation process:
annotator.run(
study_context="Adult human brain tissue samples from healthy controls and Alzheimer's disease patients, analyzed using 10X Genomics single-cell RNA-seq. Samples include cortical and hippocampal regions.",
metadata={
'experiment_name': 'Brain_AD_Study',
'run_label': 'initial_analysis'
},
save_query=True,
query_filename="query.json",
show_progress=True,
)The CyteType API provides access to some chosen LLM providers by default. Users can choose to provide their own LLM models and model providers. Many models can be provided simultaneously, and then they will be used iteratively for each of the clusters.
adata = annotator.run(
study_context="Human PBMC from COVID-19 patients",
llm_configs=[{
'provider': 'openai',
'name': 'gpt-4o-mini',
'apiKey': 'your-api-key',
'baseUrl': 'https://api.openai.com/v1', # Optional
'modelSettings': { # Optional
'temperature': 0.0,
'max_tokens': 4096
}
}],
)If you do not provide your own model providers, then the CyteType API implements rate limiting for fair usage:
- Annotation submissions: 5 RPD
- Reannotation: 10 RPD
- Report retrieval: 20 RPM
If you exceed rate limits, the system will return appropriate error messages with retry timing information
Supported providers: openai, anthropic, google, xai, groq, mistral, openrouter, bedrock
The CyteType API supports Ollama models as well. You will need to expose your Ollama server to the internet using a tunneling service. Refer to the OLLAMA.md file for instructions on how to do this.
adata = annotator.run(
...
# API polling and timeout settings
poll_interval_seconds=30, # How often to check for results (default)
timeout_seconds=7200, # Max wait time (default: 2 hours)
# API configuration
api_url="https://custom-api.com", # Custom API endpoint
auth_token="your-auth-token", # Authentication token
)You can provide your own token to the run method using the auth_token parameter. This will be included in the Authorization header as "Bearer {auth_token}". All API requests will be authenticated with this token.
CyteType performs comprehensive cell type annotation through an automated pipeline:
- Automated Annotation: Identifies likely cell types for each cluster based on marker genes
- Ontology Mapping: Maps identified cell types to Cell Ontology terms (e.g.,
CL_0000127) - Review & Justification: Analyzes supporting/conflicting markers and assesses confidence
- Literature Search: Searches for relevant literature to support the annotation
CyteType generates detailed contextual information to inform annotations:
Dataset-Level Context: Comprehensive analysis of experimental metadata:
"This dataset originates from multiple human tissues including adrenal gland,
brain, liver, lung, lymph node, and pleural effusion, with samples derived
from both healthy individuals and patients diagnosed with lung adenocarcinoma
or small cell lung carcinoma. Experimental data was generated via 10X Genomics
Chromium 3' single-cell sequencing, which may introduce platform-specific
technical artifacts."
Cluster-Specific Context: Detailed metadata analysis for each cluster:
"Cluster 1 comprises 99% lung-derived cells, with 65% originating from lung
adenocarcinoma samples and 33% from normal tissue. The cells are distributed
across two primary donors with demographic characteristics including 67%
female donors and 97% self-reported European ethnicity. Treatment conditions
include Platinum Doublet (55%) and Naive (44%)."
This contextual information enables more accurate annotations by considering:
- Tissue Origins: Multi-tissue datasets with precise anatomical mapping
- Disease States: Healthy vs. pathological conditions with treatment history
- Technical Factors: Sequencing platforms, batch effects, and processing methods
- Demographics: Age, sex, and ethnicity distributions
- Treatment Context: Therapeutic interventions and their potential cellular effects
Results include comprehensive annotations for each cluster with expert-level analysis:
# Access results after annotation using the helper method
results = annotator.get_results()
# Or access directly from the stored JSON string
import json
results = json.loads(adata.uns['cytetype_results']['result'])
# Each annotation includes comprehensive information:
for annotation in results['annotations']:
print(f"Cluster: {annotation['clusterId']}")
print(f"Cell Type: {annotation['annotation']}")
print(f"Granular Annotation: {annotation['granularAnnotation']}")
print(f"Cell State: {annotation['cellState']}")
print(f"Confidence: {annotation['confidence']}")
print(f"Ontology Term: {annotation['ontologyTerm']}")
print(f"Is Heterogeneous: {annotation['isHeterogeneous']}")
# Supporting evidence and conflicts
print(f"Supporting Markers: {annotation['supportingMarkers']}")
print(f"Conflicting Markers: {annotation['conflictingMarkers']}")
print(f"Missing Expression: {annotation['missingExpression']}")
print(f"Unexpected Expression: {annotation['unexpectedExpression']}")
# Expert review and justification
print(f"Justification: {annotation['justification']}")
print(f"Review Comments: {annotation['reviewComments']}")
print(f"Feedback: {annotation['feedback']}")
# Similarity and literature support
print(f"Similar Clusters: {annotation['similarity']}")
print(f"Corroborating Papers: {len(annotation['corroboratingPapers']['papers'])} papers")
# Model usage and performance metrics
print(f"Models Used: {annotation['llmModels']}")
print(f"Total Processing Time: {annotation['usageInfo']['total_runtime_seconds']:.1f}s")
print(f"Total Tokens: {annotation['usageInfo']['total_tokens']}")Expert Review System: Each annotation undergoes multi-stage review with detailed feedback:
- Review Comments: Expert-level biological interpretation and mechanistic insights
- Confidence Assessment: Moderate/High confidence based on marker evidence
- Feedback Loop: Iterative refinement based on biological plausibility
- Mechanistic Analysis: Discussion of signaling pathways, developmental biology, and disease pathogenesis
Literature Integration: Automatic literature search provides supporting evidence:
- Corroborating Papers: Relevant publications with PMIDs and summaries
- Biological Context: Integration of current research to validate annotations
Example corroborating papers:
papers = annotation['corroboratingPapers']['papers']
for paper in papers:
print(f"Title: {paper['title']}")
print(f"PMID: {paper['pmid']}")
print(f"Journal: {paper['journal']} ({paper['year']})")
print(f"Summary: {paper['summary']}")Sample output:
Title: YAP regulates alveolar epithelial cell differentiation and AGER via NFIB/KLF5/NKX2-1
PMID: 34466790
Journal: iScience (2021)
Summary: Documents atypical HOPX+AGER+SFTPC+ 'dual-positive' alveolar cells that
persist in mature lungs, directly validating the mixed AT1/AT2 phenotype observed
in malignant clusters.
Marker Analysis: Comprehensive evaluation of gene expression patterns:
- Supporting Markers: Genes that strongly support the annotation
- Conflicting Markers: Genes that challenge the annotation with explanations
- Missing/Unexpected Expression: Detailed analysis of expression anomalies with biological explanations
Example unexpected expression analysis:
"Expression of AT2 markers (SFTPC, SFTPB) and club cell marker (SCGB1A1)
in a cluster with strong AT1 markers"
→ Explained by: "dedifferentiation process in cancer where transformed
epithelial cells exhibit aberrant co-expression of markers from multiple
lineages due to pathological plasticity"
Performance Metrics: Detailed usage statistics for transparency:
- Model Information: Which LLM models were used for each analysis step
- Runtime Statistics: Processing time and token usage per cluster
- Annotation Attempts: Number of refinement iterations
CyteType provides sophisticated, multi-layered annotations:
Basic Cell Type: "B-cell", "Lung Adenocarcinoma Cell"
Granular Annotations: Detailed phenotypic descriptions:
"AGER-positive, HOPX-positive, KRT19-positive lung adenocarcinoma cell with mixed AT1/AT2 phenotype""EMT-transitioned, pleural metastasis-competent adenocarcinoma cell with platinum-induced stress phenotype""CD74-high activated tumor-infiltrating B-cell in lung adenocarcinoma microenvironment"
Cell States: Functional and pathological states:
"Transformed","Malignant","Activated","EMT-transitioned and stressed"
Expert Review Comments: Detailed mechanistic insights:
"The mixed AT1/AT2 phenotype observed in this malignant cluster exemplifies
the pathological dedifferentiation characteristic of lung adenocarcinoma,
but the degree of lineage promiscuity suggests exceptional cellular plasticity
beyond typical adenocarcinoma patterns. This may indicate activation of
primitive developmental pathways like Wnt/β-catenin signaling..."
git clone https://github.com/NygenAnalytics/cytetype.git
cd cytetype
uv sync --all-extras
uv run pip install -e .The package defines several custom exceptions for different error scenarios:
CyteTypeError: Base exception class for all CyteType-related errorsCyteTypeAPIError: Raised for errors during API communication (network issues, invalid responses)CyteTypeTimeoutError: Raised when API requests timeoutCyteTypeJobError: Raised when the API reports an error for a specific job
uv run pytest # Run tests
uv run ruff check . # Linting
uv run ruff format . # Formatting
uv run mypy . # Type checkingLicensed under CC BY-NC-SA 4.0 - see LICENSE for details.
