/bavaria

Batch-adversarial variational auto-encoder (BAVARIA) for simultaneous dimensionality reduction and integration of single-cell ATAC-seq datasets

Primary LanguageJupyter NotebookOtherNOASSERTION

BAVARIA

BAVARIA is python package that implements a Batch-adversarial Variational auto-encoder with Negative Multinomial reconstruction loss for single-cell ATAC-seq analysis.

In particular, the model can be used to extract a latent feature representation of a cell which can be used for downstream analysis tasks, including cell cluster, cell identification, etc. The package is freely available under a GNU Lesser General Public License v3 or later (LGPLv3+)

Installation

You can install the package version v0.1.0 via

pip install https://github.com/BIMSBbioinfo/bavaria/archive/v0.1.0.zip

Alternatively, you can install the latest version from the master branch using

pip install git+https://github.com/BIMSBbioinfo/bavaria.git

Documentation

BAVARIA offers a command line interface that fits an ensemble of BAVARIA models given a raw count matrix (-data) Subsequently, the model parameters and latent features are stored in the output directory (-output)

bavaria -data adata.h5ad \
      -output <outputdir> \
      -epochs 200 \
      -nrepeats 10 \
      -nlatent 15 \
      -batchnames batch \
      -modelname bavaria

Additional information on available hyper-parameters are available through

bavaria -h

Tutorial

Below you find links to the tutorials. The tutorials will require jupyter and other resources which are defined in tutorial/requirements.txt. Using the requirements file you instantiate a new conda environment using

conda create --name bavaria_tutorial --file tutorial/requirements.txt
Example notebooks
Data preparation PBMC integration
Using BAVARIA to integrate PBMC data