search

DS3Lab / RosENet

35 stars 11 forks source link

Some problem about preprocessing the PDBbind 2016 dataset #9

Closed wanyueli closed 4 years ago

wanyueli commented 4 years ago

When I run the python3 -m RosENet -- preprocess refine_set, I met the problem that for some data this operation cannot generate all the features mentioned in the paper, and will affect the later voxelization part. The preprocessed results that can be voxelized as shown as follows: preprocess_6 The preprocessed results that don't contain all the features and can't be voxelized as shown as follows: preprocess_7 Is anyone meet the same problem? Why does it happen?

hhassan1 commented 4 years ago

In principle the preprocessing step generates two sets of files:

  1. ligand_*.pdbql and protein_*.pdbqt, which are used to generate the molecular descriptor features (with the HTMD library)
  2. complex_*.pdb and complex_*.attr.npz used to generate the Rosetta features.

Can you confirm if any of those files looks wrong?

wanyueli commented 4 years ago

Thanks for your reply, I have solved this problem