Biochemical networks with simulation-based estimations of dynamical properties

This datasets collection was first introduced in the article: 

Exploiting the structure of biochemical pathways to investigate dynamical properties with neural networks for graphs. (Bioinformatics 2023)

The collection contains three datasets that contain information about three dynamical properties computed on a set of 483 biochemical pathways downloaded from the BioModels database. The three dynamical properties are:

• robustness
• sensitivity
• monotonicity

The files are organized as follows:

1. The `pathways` directory contains 483 files in .dot format for each biochemical pathway downloaded from the BioModels database (May 2021), represented in Petri net format (see this article for the exact definition). The file name is the ID of the pathway in the BioModels database.
2. The other folders contain one .csv file for each property. A single .csv file contains 4 columns:
   1. `PathwayID`: the ID of the Pathway in the BioModels database
   2. `Input`: the input molecular species on which the property has been assessed
   3. `Output`: the output molecular species on which the property has been assessed
   4. `Property`: the value of the property assessed with numerical simulations on the pathway for that particular input/output species pair.
3. The `loader.py` file is an optional script that allows to use the data in python. The script requires that the libraries `networkx`, `pandas`, and `pydot` are installed in the target machine.