Question about protein-peptide docking

Question

[Acervo]

Hello!

I have a question about the protein-peptide docking result I obtained. The protein was computationally predicted and has two chains. Its active site was also computationally predicted. When I did the docking, the peptide didn't bind to the supposed active site, but to another region. Based on your experience, could you tell me why this happened? Or recommend some material that talks about it? I've been looking for material on this, but I haven't found anything that can help me.

Note: The protein was well modeled and the supposed active site was indicated in three different prediction tools.

Answer 1

[Mensur Dlakic]

I don't think you will get useful suggestions without providing considerably more information. A general answer to this question:

Based on your experience, could you tell me why this happened?

There is no magic in the docking procedure that somehow "knows" where your protein is supposed to bind. If that were the case, protein docking would be trivial. Docking is energy-based, and there are many protein-peptide complexes with similar energies. Usually it helps if the docking search is restricted to a known binding site rather than move around the whole protein.

As to what is missing: What does it mean that a protein has two chains? Homodimer? Heterodimer? How do you know that it is modelled well? How do you know that a peptide is modelled well? Did you do flexible or rigid docking?

It may help if you give all the proteins and peptides to AlphaFold and let it model the complex rather than modeling individual parts and subsequently docking. I suggest you Google AlphaFold multimer for more information.

[Acervo]

In some cases, it's not possible to provide specific details. I was looking for general answer from someone who has been through a similar situation. I apologize if I didn't make this clear in my previous message. Anyway, I appreciate your help!