To enable direct scaffold hopping of macrocyclic compounds, we constructed Macro-hop models based on Reinvent (https://github.com/MolecularAI/Reinvent). It allows direct optimisation of macrocyclic compounds using 2D and 3D scoring functions and generates macrocyclic molecules with a given number of ring atoms.
- Set up conda environment and clone the github repo
- Open a shell, and go to the repository and create the Conda environment:
$ conda env create -f environment.yaml
$ conda activate Macro-hop
$ cd reinvent_functions
$ pip install reinvent_function-0.0.8.1-py3-none-any.whl
- Cuda-enabled GPU
- Linux
Running each example results in a template file; there are templates for many running modes. Each running mode can be executed by python input.py some_running_mode.json after enabling the environment.