DIRAC (Domain Invariant Respresentation through Adversatial Calibration)

Spatially resolved integration of multi-omics with DIRAC highlights cell-specific remodeling

DIRAC is a Python package, written in PyTorch and based on Scanpy.

DIRAC is a graph neural network to integrate spatial multi-omic data into a unified domain-invariant embedding space and to automate cell-type annotation by transferring labels from reference spatial or single-cell multi-omic data.

DIRAC primarily includes two integration paradigms: vertical integration and horizontal integration, which differ in their selection of anchors. In vertical integration, multiple data modalities from the same cells are jointly analyzed, using cell correspondences in single-cell data or spot correspondences in spatial data as anchors for alignment. In horizontal integration, the same data modality from distinct groups of cells is aligned using genomic features as anchors. The best way to familiarize yourself with DIRAC is to check out our tutorial and our documentation.

For more details, please check out our publication.

Directory structure

.
├── sodirac                 # Main Python package
├── data                    # Data files
├── docs                    # Documentation files
├── environment.yaml        # Reproducible Python environment via conda
├── requirements.yaml       # Python packages required for issuing DIRAC
├── LICENSE
└── README.md

How to install DIRAC

To install DIRAC, make sure you have PyTorch and PyG installed. For more details on dependencies, refer to the environment.yml file.

Step 1: Set Up Conda Environment

conda create -n dirac-env python=3.9 r-base=4.3.1 rpy2 r-mclust r-yarrr

Step 2: Install PyTorch and PyG

Activate the environment and install PyTorch and PyG. Adjust the installation commands based on your CUDA version or choose the CPU version if necessary.

• General Installation Command

conda activate dirac-env
pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu118
pip install pyg_lib==0.3.1+pt21cu118 torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.1.0+cu118.html
pip install torch_geometric==2.3.1

• Tips for selecting the correct CUDA version
  • Run the following command to verify CUDA version:

  nvcc --version
  

  • Alternatively, use:

  nvidia-smi
  

• Modify installation commands based on CUDA
  • For CUDA 12.1

    pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu121
    pip install pyg_lib==0.3.1+pt21cu121 torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.1.0+cu121.html
    pip install torch_geometric==2.3.1
    

  • For CPU-only

    pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cpu
    pip install pyg_lib==0.3.1+pt21cpu torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.1.0+cpu.html
    pip install torch_geometric==2.3.1
    

Step 3: Install dirac from shell

    pip install sodirac

Step 4: Import DIRAC in your jupyter notebooks or/and scripts

    import sodirac as sd

Installing within a conda environment is recommended.

Usage

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