/example_molbot

Primary LanguagePython

Example training of RNN

In the directory Inputs you can find different scripts:

  1. training_chembl_001.py: Script that trains the model on a large subset of the ChEMBL data set (can be found in the folder data). The output goes in the folder outputs/transfer_chembl_001.
  2. predict_chembl_001.py: This uses the model trained by script training_chembl_001.py to generate molecules. It filters any duplicate and invalid smiles and saves the remaining ones in a file predict_chembl_001.smi.
  3. transfer_learning_001.py: This performs transfer learning on the model trained with training_chembl_001.py. The transfer learning data set is metap2.smi. The model obtained is saved in tl_model.h5.
  4. predict_tl_001.py: This uses the model trained with transfer_learning_001.py and uses it to generate SMILES. It filters any duplicate and invalid smiles and saves the remaining ones in a file in predict_tl_001.smi.

In each output folder there is a *.out file which shows the standard output for each script that runs. This gives an idea of the runnning time of each script.