search

YuzheWangPKU / DiffPepBuilder

Official repository for Target-Specific De Novo Peptide Binder Design with DiffPepBuilder
MIT License
59 stars 7 forks source link

Specify some amino acids of binder #8

Open xtzhang0216 opened 2 months ago

xtzhang0216 commented 2 months ago

hi, this is a wonderful work! I'm wondering if I can fix part of the peptide. For example, I want the generated peptides to be SDRXXXXDDG, and I want the model to help me design the X residue. If not, do you have any suggestions for me to realize this requirement? thanks in advance

YuzheWangPKU commented 1 month ago

Thank you for your support of our work and sorry for the late response! Currently, our model does not support peptide inpainting, but we are actively working on integrating this functionality into the upcoming version. To the best of our knowledge, there are no existing deep learning methods that specifically focus on this task. You might consider using ProteinMPNN if the native/reference protein-peptide complex structure is available. Alternatively, you could try manually modifying the PepGenDataset class in experiments/run_inference.py file in our model to set specific residues for the peptide. This approach could potentially guide the model to preserve the fixed parts of the sequence (not tested previously).