supervised-grns

Supervised inference of gene regulatory networks from single-cell RNA-sequencing data

Installation

Download the code on your machine

git clone https://github.com/Murali-group/supervised-grns.git

Setting up conda environment and dependencies

We recommend using an Anaconda environment to run the pipeline. The below steps can be followed on your Ubuntu machine to get the pipeline setup and running.

Install conda and activate the environment

• Install Anaconda from the official website here
• After you have Anaconda installed on your machine, we will create a conda environment specific for this project. This project works on Python3

  conda create -n "SGRN" python=3.7.10 ipython

  • This will create an environment called SGRN with Python 3.7.10
  • Activate the environment with the command

    conda activate SGRN

  • We will now install the required packages and dependencies in the SGRN environment.

Install dependencies

• supervised-grns pipeline mainly uses PyTorch framework for its computation, amongst other libraries. We found that installing PyTorch can be tricky so we split this step into two -

  • Install PyTorch and related modules
  • Install rest of the packages through pip install requirements.txt

• In this step, we will install PyTorch library along with the necessary dependencies.

  • Install PyTorch in your environment with the following command - conda install -c pytorch pytorch=1.8.0
  • Check if PyTorch is installed and check the version. You should get an output like this -

    python -c "import torch; print(torch.__version__)"
    1.8.0

  • Similarly, install torchvision and check if is correctly installed -

    conda install -c pytorch torchvision
    python -c "import torchvision; print(torchvision.__version__)"
    0.9.0

  • Finally, the PyTorch geometric library can be installed by following the steps here listed under Installation via Binaries

• For the evaluation, we additionally need R packages. For this install R within the SGRN conda environment using:

  • conda install r=3.5.0
  • R -e "install.packages('https://cran.r-project.org/src/contrib/PRROC_1.3.1.tar.gz', type = 'source')"

• We now install the rest of the libraries using the requirements.txt file.

  pip install -r requirements.txt

That's it! We can now run the pipeline and check if everything is working fine.

Run the pipeline

Navigate to your folder where the code is downloaded. Run the following command to run the pipeline.

 python main.py --config=config/config.yaml 

The output is stored under <output_dir>/<output_prefix>/ (the parameter values are obtained from output_settings in the config.yaml). The output folder contains rankedEdges.csv files with the edge list for each algorithm for each random seed and a statsperFold.csv and statsAggregated.csv files containing the evaluation results stored in a tab-separated file.