about training

[csc1233]

Hello,
I would like to run the training code and noticed that it requires a dataset in .pkl format. Does this dataset need to be processed with the two pre-trained weights before it can be used as input for training? Could you please provide a script to convert a CSV file to a .pkl file that can be used for training?
Thank you for your help!

[zchwang]

The pkl file contains enzyme and substrate features. The scripts for generating these features can be found at https://github.com/zchwang/CataPro/blob/master/inference/utils.py. You can save these features into a pkl file using pandas.