Missing source files and pretrained models for reproducing AMG

[ltl888]

Dear AMG authors,

Thank you for developing AMG and making the code publicly available.

I am trying to reproduce the experiments according to your paper:

"Deep reinforcement learning as an interaction agent to steer fragment-based 3D molecular generation for protein pockets"

The paper states in the Code availability section that:

"The source code and pre-trained models for this study are available at https://github.com/ispc-lab/AMG."

However, after downloading the current GitHub repository and setting up the environment, I found that several files required by the provided scripts appear to be missing.

For example, running:

python scripts/sample_testset.py
--config configs/rl.yml
--start_index 0
--end_index 0

results in:

ModuleNotFoundError: No module named 'models.bio_markov'

I also found that:

• models/common.py is missing, although models/encoders/schnet.py imports:
  from ..common import GaussianSmearing, ShiftedSoftplus

• Several utility files seem absent:
  utils/misc.py
  utils/datasets.py
  utils/data.py
  utils/train.py
  utils/mol_tree.py
  utils/transforms.py
  utils/metric.py

• The pretrained checkpoints mentioned for inference are also not included:
  ckpts/pr_model.pt
  ckpts/rl_agent.pt

Could you please confirm whether there is a complete version of the repository or an additional package containing the missing source files and pretrained models?

I would greatly appreciate any guidance on reproducing the reported results.

Thank you very much for your help.

Best regards,