cellproportion 1.1.1

Cell type proportion analysis for single-cell and spatial transcriptomics data

CellProportion

A Python package for comparing cell type proportions between experimental groups in single-cell RNA-seq and spatial transcriptomics data.

🚀 Features

• Flexible Input: Works directly with AnnData objects or Pandas DataFrames
• Multiple Statistical Methods:
  • signed_r2 – Signed R² from regression (captures direction + fit strength)
  • mean_diff – Simple difference in mean proportions
  • log2_fc – Log₂ fold-change for multiplicative differences
  • corr – Pearson correlation with group labels
• Spatial Analysis: Compare proportions within tissue regions/spatial domains
• Statistical Testing: Mann-Whitney U tests with significance categorization
• Visualization Ready: Built-in colour schemes and customizable colour mapping
• Robust Error Handling: Comprehensive validation and informative warnings

📦 Installation

pip install cellproportion

🔧 Quick Start

Single-Cell Analysis

import pandas as pd
import scanpy as sc
from cellproportion import cell_type_abundance

# Load your single-cell data
adata = sc.read_h5ad("your_data.h5ad")

# Compare cell type proportions between conditions
results = cell_type_abundance(
    adata,                          # AnnData object
    annotation="cell_type",         # Column with cell type labels
    sample_types="condition",       # Column with experimental conditions
    sample_ID="patient_id",         # Column with sample/patient IDs
    sample_types_1="tumor",         # First group for comparison
    sample_types_2="normal",        # Second group for comparison
    method="signed_r2",             # Statistical method
    signed_r2_cutoff=0.15,         # Optional: cutoff for significance
    explain=True                    # Print method explanation
)

print(results.head())

Spatial Transcriptomics Analysis

from cellproportion.spatial import spatial_cell_type_abundance

# Analyze proportions within each spatial region
spatial_results = spatial_cell_type_abundance(
    adata,                          # AnnData with spatial information
    region_col="tissue_region",     # Column with spatial region labels
    annotation="cell_type",
    sample_types="condition", 
    sample_ID="patient_id",
    sample_types_1="tumor",
    sample_types_2="normal",
    method="signed_r2"
)

print(f"Analyzed {spatial_results['region'].nunique()} spatial regions")
print(spatial_results.head())

Using DataFrames

# Works with any DataFrame containing the required columns
metadata_df = pd.DataFrame({
    'cell_type': ['T_cell', 'B_cell', 'Macrophage'] * 100,
    'condition': ['tumor', 'normal'] * 150,
    'patient_id': ['P1', 'P2', 'P3'] * 100,
    # ... other columns
})

results = cell_type_abundance(
    metadata_df,
    annotation="cell_type",
    sample_types="condition",
    sample_ID="patient_id",
    sample_types_1="tumor",
    sample_types_2="normal"
)

📊 Understanding the Results

The output DataFrame contains:

• anno: Cell type annotation
• stat_values: Statistical metric value (depends on method chosen)
• p_values: P-value from Mann-Whitney U test
• sig_p: Significance category (p<0.01, p<0.05, p<0.1, p<0.5, p>0.5)
• colour: Colour code for visualization

# Example output
print(results[['anno', 'stat_values', 'p_values', 'sig_p']].head())
#      anno   stat_values   p_values   sig_p
# 0   T_cell  0.234567  0.0123  p<0.05
# 1   B_cell -0.123456  0.2341  p<0.5
# 2  NK_cell  0.456789  0.0001  p<0.01

🎨 Custom Colour Mapping

Create a TSV file with your preferred colours:

annotation	colour
T_cell	#E41A1C
B_cell	#377EB8
NK_cell	#4DAF4A
Macrophage	#984EA3

results = cell_type_abundance(
    adata,
    # ... other parameters
    colours_file="my_colours.tsv"
)

📈 Statistical Methods Explained

from cellproportion.methods import explain_stat_values

# Get detailed explanations of all methods
explanations = explain_stat_values()
for method, info in explanations.items():
    print(f"\n{method.upper()}:")
    for key, value in info.items():
        print(f"  {key}: {value}")

Method Comparison

Method	Best For	Pros	Cons
signed_r2	Linear relationships	Direction + fit strength	Assumes linearity
mean_diff	Simple comparisons	Easy interpretation	Ignores variance
log2_fc	Multiplicative changes	Ratio-based	Sensitive to low values
corr	Association strength	Scale-invariant	Only linear association

🔬 Advanced Usage

Batch Processing Multiple Datasets

datasets = ["dataset1.h5ad", "dataset2.h5ad", "dataset3.h5ad"]
all_results = []

for dataset_path in datasets:
    adata = sc.read_h5ad(dataset_path)
    results = cell_type_abundance(adata, method="signed_r2")
    results['dataset'] = dataset_path
    all_results.append(results)

combined_results = pd.concat(all_results, ignore_index=True)

Method Comparison

methods = ["signed_r2", "mean_diff", "log2_fc", "corr"]
method_comparison = {}

for method in methods:
    results = cell_type_abundance(adata, method=method)
    method_comparison[method] = results

# Compare results across methods
comparison_df = pd.DataFrame({
    method: method_comparison[method].set_index('anno')['stat_values'] 
    for method in methods
})

📝 Citation

If you use CellProportion in your research, please cite:

@software{cellproportion2024,
  author = {Patel, Ankit},
  title = {CellProportion: Cell type proportion analysis for single-cell and spatial transcriptomics},
  url = {https://github.com/avpatel18/cellproportion},
  version = {1.0.0},
  year = {2025}
}

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🐛 Bug Reports

If you encounter any bugs or have feature requests, please file an issue on GitHub Issues.

📧 Contact

• Author: Ankit Patel
• Email: [e-mail]
• GitHub: @avpatel18