DeepRank core generally allows you to predict 2 different things: protein-protein-interactions (PPIs) or something about variant residues.

All documentation in README is focussed on PPIs. It is unclear what the purpose of the Variant queries is and how to use this. We should add an explanation of this to the README.

Also, development of the package is done (almost) exclusively with PPIs in mind, given that it is unclear what the purpose of the variant queries is. It would really help if this was clarified so that both types of projects can be taken into account during development.