how to input a new patent (such as .csv file) and then use the trained model to predict and rank the compounds in new patent

[MachineGUN001]

hi，

Many thanks for offering such interesting and useful work.

Can you please provide a tutorial on how to generate the csv file (like the mini-test.csv) as input file for a new patent (such as .csv file) and then use the trained model to predict and rank the compounds in new patent?
how can I generate the input (test) csv file with 3272 features?
Is there any other code or script that can implement the feature generation as the paper described the workflow?
and what is the P_Ca_SMILES field in the patent csv file?
So sorry for those naive questions.

many many thanks,

Best,
Sh-Y