search

HySonLab / Ligand_Generation

Target-aware Variational Auto-encoders for Ligand Generation with Multimodal Protein Representation Learning
https://doi.org/10.1088/2632-2153/ad3ee4
26 stars 3 forks source link

pre-trained model #4

Open msako82 opened 7 months ago

msako82 commented 7 months ago

I'm very interested in your job. I downloaded the source code, but I can not download the pre-trained unconditional VAEs model. The file does not exist. Could you please upload it again?

Or, I would like to sample ligand for a specific PDB file, can you provide a learned model?

nnkhang19 commented 7 months ago

Hi,

Thanks for reaching out. We apologize for your inconvenience. As we have updated the code with more experiments on the PDBBind dataset, I just uploaded several new files and checkpoints of our models. Please check the checkpoint folder, where you can find some checkpoints, including the pre-trained unconditional VAEs and the fine-tuned model on the PDBBind dataset (target-aware.pt). Additionally, the pdb files used in our study are stored in the pdb folder.

msako82 commented 7 months ago

Thank you so much for your job! I will try to use target-aware.pt. In the generate_with_specific_target.py code, checkpoint file is set to "checkpoint/target_aware.pt". Is it correct that the ligand sampling can be done with fine tuned models in the PDBBind dataset by using the following? python3 generate_specific_target.py --protein_name [target name]

msako82 commented 7 months ago

I tried to use generate_specific_target.py, but I can not import "get_esm". I looked for how to install "get_esm" but could not find it. Is this your own function? Do you have the source code?

nnkhang19 commented 5 months ago

Hi @msako82, I apologize for the late response. I had some deadlines to finish in April and May. Now, I just uploaded the "get_esm.py" to the repository. Now, you can rerun the code with this file. Thank you for considering our work. Please let me know if there are any further issues.