Status: Manuscript under review, codebase released
This is the codebase for the manuscript Developing A Novel Causal Inference Algorithm for Personalized Biomedical Causal Graph Learning Using Meta Machine Learning by Hang Wu, Wenqi Shi, and May Dongmei Wang.
Below we include instructions for running the algorithms proposed in the manuscript.
Complete the following steps:
- Clone the repo to your desired locations
- Create a virtual python environment using a environment management software of your choice (e.g. conda, venv)
- Install the prerequistes via
pip install -r requirements.txt
Note: this code is tested under python3.9 and pytorch build 2.2.2 with CUDA11.8. This specific version could be installed via command pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
.
To run a model, simply run python main.py
and you will obtain relevant code outputs.
The dataset generation and loading scripts can be found in the data folder. And the SACHS dataset can be downloaded in the link.
We thank the authors of the paper DAGs with NO TEARS: Continuous Optimization for Structure Learning for releasing their codes on GitHub, and for simplicity of usage, we import their codebases into our folder structure as-is and build on top of it.