A scope selection tool for organic chemistry that balances scope performance and substrate similarity. Check out our preprint for more information!
-
Get scope substrate suggestions
-
Generate molecular features and the search space input for the algorithm
-
Visualize the scope selections on a UMAP
-
SHAP feature analysis
-
Predictive modeling for unseen substrates outside of the scope
-
ScopeBO is also available as an app (see below)
(1) Install Miniconda.
Note: We noticed that Anaconda can sometimes lead to installation problems due to using a different environment solver.
(2) Download this repository.
(3) Open the terminal in the ScopeBO folder (containing the file "environment.yml").
(4) Run the following command to install the environment:
conda env create -f environment.yml
(5) Activate the environment by running the following command:
conda activate scope_bo
(6) Launch Juypter Notebook by running the following command:
jupyter notebook
Open the file ScopeBO_example.ipynb in the folder "Examples" to see a usage example.
See the file installation-instructions.txt for more detailed instructions.
(1) Before first use: Follow the installation instructions steps 1–4 (see above).
(2) Double-click the app (ScopeBO_App.zip on MacOS or ScopeBO_App.bat on Windows).
Note: Unpack the zip folder on MacOS by double-clicking and then open the unpacked ScopeBO_App.app.
Note: See the file installation-instructions.txt for information on opening the app via the terminal or in case of security settings blocking the opening of the app.
-
See the supporting information of our publication for detailed recommendations regarding search space curation, featurization, etc. This information is also provided in the app via the "?" help buttons.
-
Publication data for our manuscript can be found here.