search

choderalab / espaloma

Extensible Surrogate Potential of Ab initio Learned and Optimized by Message-passing Algorithm 🍹https://arxiv.org/abs/2010.01196
https://docs.espaloma.org/en/latest/
MIT License
212 stars 23 forks source link

[Question] Retraining espaloma model to include nonbonded potentials #125

Closed nbruciaferri closed 1 year ago

nbruciaferri commented 2 years ago

Hello, I'm trying to retrain espaloma model on my own data and I was wondering how can I extract non bonded potentials. As far as I see in the GitHub repo the code that includes non bonded calculations in espaloma/mm/energy.py the application of nonbonded energies is commented out. So my questions are:

  1. How can I do to retrain the model including also non bonded energies calculation?
  2. How can I then extract them from the graph after applying the model to a new molecule?

Thanks!

yuanqing-wang commented 1 year ago

hi @niccolo93 currently we haven't tested including non-bonded energies in the training yet. but you can define any function forms you wish as long as the features you specify in JanossyPooling https://github.com/choderalab/espaloma/blob/ec45de310836cede8776b426d64d54aad9f9efd1/espaloma/nn/readout/janossy.py#L12 matches the energy functions you write.