Python interface to disease-related cell atlas approximations

atlasapprox_disease is a Python interface designed to provide easy access to disease-related cell atlas data. Built using the cellxgene census as the initial data source and powered by the scquill approximation algorithm, this package enables researchers to explore thousands of datasets efficiently.

The package enables researchers to address complex biological questions across multiple organs, cell types, and disease conditions such as:

• What is the expression of a specific gene in a specific disease across all datasets?
• In COVID-19, what is the differential cell abundance of each cell type between normal and disease states?
• What are the top 20 most differentially expressed genes in kidney disease?

Features

• Differential analysis:
  • Compare differential gene expression across diseases, tissues, sexes, or conditions.
  • Retrieve differential cell type abundance between disease states.
• Expression & Abundance Queries::
  • Retrieve average gene expression levels across multiple datasets
  • Compute the fraction of cells expressing specific genes.
  • Identify the highest-expressing cell types for a given gene across diseases and datasets.
• Metadata Access:
  • Retrieve dataset metadata, including cell types, tissues, diseases, sex, development stage.
• Visualization::
  • Generate dot plots to visualize gene expression and detection frequencies.

Installation

To install the package, run:

pip install atlasapprox-disease

Quick start

Initialise the API

import atlasapprox_disease as aad
api = aad.API()

Retrieve Metadata

metadata = api.metadata()
print(metadata.head())

Query differential gene expression

diff_exp = api.differential_gene_expression(
    differential_axis="disease",
    disease="Diabetes",
    top_n=10,  # Get top 10 upregulated & downregulated genes
    method="delta_fraction"
)
print(diff_exp)

Generate average gene expression

average_exp = api.average(
    features="INS,GCK,MAFA,PECAM1",
    disease="Diabetes",
    include_normal=True
)
print(average_exp)

For more detailed tutorials and examples, please visit the API github repo.