theia
theia copied to clipboard
Code on explaining contributions
I am kindly asking where is the code for explaining the contributions (from SMILES to structures).
In the paper: "By mapping the molecular fragments occurring in the reactions to the vector entries that act as input for a neural network classifier, we enable the use of the DeepLIFT algorithm (implemented as DeepSHAP) to annotate fragments and atoms with their respective classification contributions, providing chemical explainability for all described models (Fig. 1(1))."
Best regards, Haoke Qiu