Octavian Ganea

Hi, Yes, you are right, the DB5.5 data contains many inconsistencies between the bound and unbound structures for the same protein, not sure exactly why. To generate that plot, we...

Yep, here is the code (also see the full list of RMSD computed below). ``` import os import numpy as np from src.utils.protein_utils import rigid_transform_Kabsch_3D from src.utils.db5_data import get_residues_db5 import...

Generating the full DIPS training data takes a lot of time and you have to check if you have enough resources for it. Can you try generating just a fraction...

The ligand is docked to the bound receptor, so only the final ligand file is the output. See https://github.com/octavian-ganea/equidock_public/blob/main/src/inference_rigid.py#L136 . The initial bound ligand is not needed for inference alone,...

This looks like a DGL error. Do you have the same package versions as stated in our README ?

all packages need to be the same versions , not just DGL. Can you show your packages versions for those listed in the README ?

also, fyi, this code was not tested on Windows

it really looks to me like a DGL error. Can you try testing the hetero_graph.update_all() function in a separate toy code ?

I replaced this complicated SplitCrossEntropyLoss by the pytorch cross entropy loss which produces the same results and seems to be only slightly slower.

@RicardoUsbeck I have the Wiki annotations which are separated from the original dataset, but it's best if you can ask for them directly to the authors of this paper: http://www.aclweb.org/anthology/W10-3503...