README.md

Transformer for Molecular Property Prediction: Lessons Learned from the Past Five Years

This repo provides the data and the code used to generate the figures and tables in the corresponding review article:

• Arxiv

Project requirement

The project is compatible with Python version >= 3.11, and the packages are managed by poetry. So, you need to have poetry installed on your device. You can do so by running this command

curl -sSL https://install.python-poetry.org | python3 -

For more information, refer to the documentation page

To install the project dependencies, run

poetry install --no-root

Execution

To reproduce the figures and the tables, simply run

python3 main.py

The main.py file calls the corresponding functions from the scripts folder. These functions access their data from the data folder. The figures and tables will be saved to the locations as shown in the printed messages after excution.

You can also check the reproduce_figures.ipynb file for visualising the figures directly.

Supplementary info

In table 6, we mention that some of the information are extracted from the provided code of the papers as the info was not stated in the original manuscript. The permalink for each of these values are provided at the data/num_parameters/info_extracted_from_code.csvfile.