ajalamarvellous/graph-eg5

graph-eg5

Application of graph neural networks [GCN & GAT] for the application of natural compounds that inhibit eg5 target

This repository contains the entire scientific project, including code and report. The philosophy behind this repository is that no intermediary results are included, but all results are computed from raw data and code.

Getting ready

You need mamba to run the analysis. Using mamba, you can create an environment from within you can run it:

mamba env create -f environment.yaml --no-default-packages

Run the analysis

snakemake

This will run all analysis steps to reproduce results and eventually build the report.

You can also run certain parts only by using other snakemake rules; to get a list of all rules run snakemake --list.

To generate a PDF of the dependency graph of all steps build/dag.pdf run:

snakemake dag

Be notified of build successes or fails

As the execution of this workflow may take a while, you can be notified whenever the execution terminates either successfully or unsuccessfully. Notifications are sent by email. To activate notifications, add the email address of the recipient to the configuration key email. You can add the key to your configuration file, or you can run the workflow the following way to receive notifications:

  snakemake --config email=<your-email>

Run the tests

snakemake test

Repo structure

• report: contains all files necessary to build the report; plots and result files are not in here but generated automatically
• scripts: contains the Python source code as scripts
• rules: contains Snakemake rule definitions
• envs: contains execution environments
• tests: contains the test code
• config: configurations used in the study
• profiles: Snakemake execution profiles
• data: place for raw data
• build: will contain all results (does not exist initially)

License

The code in this repo is MIT licensed, see ./LICENSE.md.