Hi! I’m currently trying to see how Prodigy extracted the parameters from training data to do calculations. Is there training code I can view to see how Prodigy was made? I’m trying to figure out how to get it to work for protein-aptamer complexes in addition to protein-protein, so I was wondering if the training code is available.
The training was made with a rather standard package in R (a regression model optimizing weights)
I would think you could use any package for that.
The git repo does allow you to calculate the contacts (type and number), which are used by the model.
Sounds good, thank you for the reply! Do you have the R code you used to get the weights for Prodigy? Can you send me the link? What are some resources I can use to have R get the parameters for protein-aptamer complexes to adapt Prodigy to get Kd values for protein-aptamer?
The training was done more than 10 years ago…
Doing some archeology to find the R scripts doesn’t make much sense.
I would suggest to use rather modern libraries for this, either in R or python (e.g. https://scikit-learn.org/stable/)