scelvis 0.1.0

Installation

The only prerequisite is Python 3, everything else will be installed together with the scelvis package.

You can install SCelVis and its dependencies using pip or through conda:

$ pip install scelvis
# OR
$ conda install scelvis

A Docker container is also available:

$ docker run bihealth/scelvis:latest --help
$ docker run -p 8050:8050 -v data:/data bihealth/scelvis:latest run --data-dir /data

Preparing Your Data

Each data set consists of an HDF5 file called data.h5ad and a dataset description file about.md. The HDF5 file is an anndata object that stores gene expression (sparse CSR matrix) and meta data with very fast read access. You can use the scelvis convert command for converting your single-cell pipeline output into an appropriate HDF5 file. The about.md file should look as follows:

----
title: An Optional Long Data Set Title
short_title: optional short title
----

A verbose description of the data in Markdown format.

A directory containing both an data.h5ad and an about.md file is a dataset directory. For the input you can either specify one dataset directory or a data directory containing multiple dataset directories.

You can convert your single-cell transcriptome analysis pipeline as follows. This does no further processing except log-normalization and uses PCA, tSNE, and clustering performed by cellranger

$ mkdir -p data/project
$ scelvis convert --input-dir cellranger-out --output-dir data/project
$ cat <<EOF
----
title: My Project
----

This is my project data.
EOF

$ tree data
data
├── other
│   ├── about.md
│   └── data.h5ad
└── project
    ├── about.md
    └── data.h5ad

Note that right now only CellRanger output is supported.

Visualizing Your Data

$ tree data
data
├── other
│   ├── about.md
│   └── data.h5ad
└── project
    ├── about.md
    └── data.h5ad

$ scelvis run --data-dir data/project
# OR
$ scelvis run --data-dir data

Developer Setup

The prerequisites are:

• Python 3, either

  • system-wide installation with virtualenv, or

  • installed with Conda.

For virtualenv, first create a virtual environment and activate it.

$ virtualenv -p venv
$ source venv/bin/activate

For a Conda-based setup create a new environment and activate it.

$ conda create -y -n scelvis 'python>=3.6'
$ conda activate scelvis

Next, clone the repository and install the software as editable (-e). Also install the development requirements to get helpers such as black.

$ git clone git@github.com:bihealth/scelvis.git
$ cd scelvis
$ pip install -e .
$ pip install -r requirements/develop.txt

Afterwards, you can run the visualization web server as follows:

$ scelvis run --data-dir path/to/data/dir