Solvent-MLP

Use of Conda and Dependencies

If one wishes to replicate the MLP training or plotting, a Conda enviornment is provided and can be used by the comand:

conda env create -f solvent.yml

Scripts can then be run in Python once Conda has been downloaded and the solvent enviornment has been created.

##External Data Multiple files are hosted externally due to size limitations. For accessing the parsed USPTO reaction SMILES, the Daniel Lowe figshare repo should be used.

Training

If one wishes to replicate the k-fold training of MLPs, a script is provided for such. This should be run from the root directory from this repo as:

python training-scripts/k-fold-training.py

Plotting

If one wishes to replicate plotting results they should again be run from the root directory.