scglue 0.3.2

Graph-linked unified embedding for unpaired single-cell multi-omics data integration

GLUE (Graph-Linked Unified Embedding)

Graph-linked unified embedding for single-cell multi-omics data integration

For more details, please check out our publication.

Directory structure

.
├── scglue                  # Main Python package
├── data                    # Data files
├── evaluation              # Method evaluation pipelines
├── experiments             # Experiments and case studies
├── tests                   # Unit tests for the Python package
├── docs                    # Documentation files
├── custom                  # Customized third-party packages
├── packrat                 # Reproducible R environment via packrat
├── env.yaml                # Reproducible Python environment via conda
├── pyproject.toml          # Python package metadata
├── LICENSE
└── README.md

Installation

The scglue package can be installed via conda using one of the following commands:

conda install -c conda-forge -c bioconda scglue  # CPU only
conda install -c conda-forge -c bioconda scglue pytorch-gpu  # With GPU support

Or, it can also be installed via pip:

pip install scglue

Installing within a conda environment is recommended.

Usage

Please checkout the documentations and tutorials at scglue.readthedocs.io.

Development

Install scglue in editable form via flit (first install flit via conda or pip if not installed already):

flit install -s

Run unit tests:

pytest --cov="scglue" --cov-report="term-missing" tests [--cpu-only]

Build documentation:

sphinx-build -b gettext docs docs/_build/gettext
sphinx-intl update -p docs/_build/gettext -l zh_CN -d docs/locale
sphinx-build -b html -D language=en docs docs/_build/html/en  # English version
sphinx-build -b html -D language=zh_CN docs docs/_build/html/zh_CN  # Chinese version

Reproduce results

1. Checkout the repository to v0.2.0:

   git checkout tags/v0.2.0
   

2. Create a local conda environment using the env.yaml file, and then install scglue:

   conda env create -p conda -f env.yaml && conda activate ./conda
   flit install -s
   

3. Set up a project-specific R environment:

   packrat::restore()  # Packrat should be automatically installed if not available.
   install.packages("data/download/Saunders-2018/DropSeq.util_2.0.tar.gz", repos = NULL)
   install.packages("custom/Seurat_4.0.2.tar.gz", lib = "packrat/custom", repos = NULL)
   

   R 4.0.2 was used during the project, but any version above 4.0.0 should be compatible.

4. Follow instructions in data to prepare the necessary data.

5. Follow instructions in evaluation for method evaluation.

6. Follow instructions in experiments for case studies.