checkatlas 0.8.0

One liner tool to check the quality of your single-cell atlases.

CheckAtlas is a one liner tool to check the quality of your single-cell atlases. For every atlas, it produces the quality control tables and figures which can be then processed by multiqc. CheckAtlas is able to check the quality of Scanpy, Seurat, and CellRanger files.

More information on the read the doc page

Summary

Powered by nextflow, checkatlas can be ran in one command line:

checkatlas run search_folder/

The checkatlas workflow start with a fast crawl through your working directory. It detects Seurat (.rds), Scanpy (.h5ad) or cellranger (.h5) atlas files.

Then, it goes through all atlas files and produce summary information:

• All basic QC (nRNA, nFeature, ratio_mito)
• General information (nbcells, nbgenes, nblayers)
• All elements in atlas files (obs, obsm, uns, var, varm)
• Reductions (pca, umap, tsne)
• All metrics (clustering, annotation, dimreduction, specificity)

All tables and figs are saved in the checkatlas_files folder in your search folder.

A single html report is produced, using MultiQC, in checkatlas_files/Checkatlas-MultiQC.html.

Examples

• Evaluate and compare different scanpy atlases: Example 1

• Evaluate different version of one atlas: Example 2

• Evaluate Scanpy, Seurat and CellRanger objects in your folder: Example 3

• Evaluate an integrated Scanpy atlas with the corresponding raw CellRanger atlases: Example 4

• Evaluate different Cellranger atlases with multiple chemistry version and cellranger version: Example 5

Installation

PyPi

The checkatlas python module can be downloaded with PyPi

pip install checkatlas

To benefit from the jax optimization of metrics you should install the version corresponding to your hardware

pip install checkatlas[cpu]        → CPU-only JAX
pip install checkatlas[cuda13]     → NVIDIA CUDA 13
pip install checkatlas[tpu]        → Google Cloud TPU

Conda

conda install bioconda::checkatlas

Docker

docker pull quay.io/biocontainers/checkatlas

R install (obsolete for now)

Finally, checkatlas comes with rpy2 to perform the interface between python and R. But, it does not automatically install Seurat. So if you want to screen Seurat atlases you need to perfrom this last installation

% R
> install.packages('Seurat')
> library(Seurat)

Development

This project is in a very early development phase. All helpers are welcome. Please contact us or submit an issue.

Read the CONTRIBUTING.md file.

Checkatlas has two repositories:

• The checkatlas module
• The checkatlas nextflow workflow

It has a module on MultiQC

• MultiQC checkatlas branch

The checkatlas package is available on PyPI

• Checkatlas PyPI

The bioconda recipe has been submitted

• Checkatlas bioconda recipe

Project developed thanks to the project template : (https://github.com/rochacbruno/python-project-template/)