search

HannesStark / EquiBind

EquiBind: geometric deep learning for fast predictions of the 3D structure in which a small molecule binds to a protein
MIT License
473 stars 109 forks source link

[Question] Graph representations #19

Closed rhjohnstone closed 2 years ago

rhjohnstone commented 2 years ago

Hi, really interesting paper. Could you please clarify something for me: when creating edges for the ligand graph, do you use both k-NN and distance threshold? Or do you use only the distance threshold for the ligand, and k-NN for the protein? Thanks!

HannesStark commented 2 years ago

Hi, For the ligand graph, we only use a distance cutoff and there is no maximum on the number of edges.