Open astomer2 opened 2 weeks ago
Oh, I have an additional question. I want to reproduce your peptide docking work. The function that needs to be referenced in train_pl.py: "from utils.dataset import ForeverTaskDataset" does not exist. I searched all the files and this module does not exist.
Hi, thanks for your interest. The reason to include Autodock Vina as a baseline of peptide docking is to explore whether current small-molecule docking tools can be immediately applied to peptide docking, by simply regarding the peptide as a special case of small molecule as PocketXMol does. We found they cannot be directly applied to peptide docking. We want to prove that although PocketXMol can dock both small molecules and peptides through a unified view, it does not mean any small-molecule docking tools can be trivially extended to peptides. Actually, we found that technical issues existed when applying those small-molecule docking tools to peptides due to their tailored design for small molecules. As for ADCP, it is a typical peptide docking baseline, in the same category as FlexPepDock and AlphaFold. We may consider adding it later.
Thanks for your reminder of the lack of files. Now we have updated the repo and improved the configuration examples for easier usage. Feel free to contact me if you have any other questions.
First of all, thank you for your excellent work. I am very interested in your ideas and tools. I have some questions when reading the peptide docking part. I know a tool specifically for peptide docking, ADCP, which is a tool based on Autodock. Your paper shows that pocketxmol is far superior to Autodock and vina in peptide docking ability. Can you show or remind us the difference between pocketxmol and ADCP in peptide docking ability?