ChimeraX protein-protein binding site modeling question
Hello, I work in a lab that researches biopesticides in the form of insect neuropeptides, which bind to receptor proteins. I am trying to use ChimeraX to model potential binding sites between peptide ligands and their corresponding receptor. Does ChimeraX have this capability? When I was trying to figure this out myself online through the ChimeraX Youtube tutorials and tutorials on the UCSF site, I came across a comment on a video that mentioned I could upload the sequences of my receptor and peptide ligand separated by a comma, and model it as a protein complex, and then I can get information on where the two molecules interact. I'm relatively new to protein research, and so I hadn't yet thought of ligand-protein interactions as a type of protein complex. My specific questions are:
* When modeling a peptide ligand and a receptor as a protein complex, will ChimeraX accurately predict the relationship between the two molecules based on binding compatibility? Or is it based on something else, like similarity of sequence alignments...?
* If multiple binding sites are possible, is there any way to view the different options for multiple binding sites? Or is there any way to restrict the area on the receptor to which my target peptide ligand is predicted to bind (i.e. the extracellular region only)?
* The peptide ligands are only 7-9 amino acids in length and do not take very long to model, but the receptor is about 600 amino acids and takes a much longer time to model. Do I have to model protein complexes by re-modeling everything for each different receptor-ligand combination, or can I combine existing models somehow?
I hope these questions aren't too troublesome, and I hope I haven't grossly misunderstood ChimeraX's capabilities.
Thank you for your time,
Ryssa Parks
Hi Ryssa, ChimeraX does not predict ligand binding sites/conformations, nor does it predict protein-protein interactions.
What you saw in the video is probably using AlphaFold Multimer. There is a tool in ChimeraX that allows running the AlphaFold Multimer calculation using Google Colab (not inside of ChimeraX, although you would get the results back in ChimeraX when the calculation finishes).
So you would have to read about the AlphaFold Multimer program and its performance on protein-protein interaction prediction to try to figure out whether it is appropriate for your situation. We are not the developers of AlphaFold Multimer.
You can see the help page for the ChimeraX "AlphaFold" tool, which includes links to the papers about AlphaFold and AlphaFold Multimer. https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html
However: It would probably be more appropriate to use a program intended for protein-peptide docking, which is exactly the problem you have: predicting how/where a peptide binds to your protein. Such programs don't predict the protein structure, they just use the known structure you give it... although they may allow some flexibility of the protein sidechains and binding site region, along with flexibility of the peptide.
I am not an expert in this area, so I don't know what the best programs are currently for protein-peptide docking. Maybe others have recommendations, or you can do some literature/web searching on that topic. My simple google search "protein-peptide docking" turned up this paper from 2020, for example: https://pubs.acs.org/doi/10.1021/acs.jctc.9b01208
... but I'm sure there are even more than 14 protein-peptide docking programs, and some of them newer than 2020.
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Department of Pharmaceutical Chemistry University of California, San Francisco
On Dec 20, 2023, at 10:06 AM, Parks, Ryssa via ChimeraX-users chimerax-users@cgl.ucsf.edu wrote:
Hello, I work in a lab that researches biopesticides in the form of insect neuropeptides, which bind to receptor proteins. I am trying to use ChimeraX to model potential binding sites between peptide ligands and their corresponding receptor. Does ChimeraX have this capability? When I was trying to figure this out myself online through the ChimeraX Youtube tutorials and tutorials on the UCSF site, I came across a comment on a video that mentioned I could upload the sequences of my receptor and peptide ligand separated by a comma, and model it as a protein complex, and then I can get information on where the two molecules interact. I’m relatively new to protein research, and so I hadn’t yet thought of ligand-protein interactions as a type of protein complex. My specific questions are:
· When modeling a peptide ligand and a receptor as a protein complex, will ChimeraX accurately predict the relationship between the two molecules based on binding compatibility? Or is it based on something else, like similarity of sequence alignments…? · If multiple binding sites are possible, is there any way to view the different options for multiple binding sites? Or is there any way to restrict the area on the receptor to which my target peptide ligand is predicted to bind (i.e. the extracellular region only)? · The peptide ligands are only 7-9 amino acids in length and do not take very long to model, but the receptor is about 600 amino acids and takes a much longer time to model. Do I have to model protein complexes by re-modeling everything for each different receptor-ligand combination, or can I combine existing models somehow?
I hope these questions aren’t too troublesome, and I hope I haven’t grossly misunderstood ChimeraX’s capabilities.
Thank you for your time,
Ryssa Parks
ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
Hi Ryssa,
I'm not sure if AlphaFold multimer will run on protein in complex with a 7-9 amino acid peptide because it does a deep sequence alignment for each of the sequences and the peptide sequence is probably too short and AlphaFold may reject it. You can just try it and see if it makes a prediction or gives an error.
If it makes a prediction then yes it is trying to produce energetically favorable binding. I've seen it succeed docking a signal peptide to a protein as shown here
https://www.rbvi.ucsf.edu/chimerax/data/alphafold-contacts-jun2022/afcontact...
and that page tells you how you can see if AlphaFold has high or low confidence in the predicted binding.
I think your case is not likely to work unless the peptide is actually part of a bigger protein and you give AlphaFold that full protein sequence. The reason is that AlphaFold looks at evolutionary patterns in the sequences by making a deep sequence alignment often containing thousands of related sequences. But a 7-9 amino acid peptide is too short to get evolutionary information by alignment to all known protein sequences. Still there is some small chance of it working because AlphaFold also has learned about plausible binding even when it does not have sequence evolution info.
You cannot specify where to try to bind the peptide with AlphaFold. It will make 5 predictions and you can load the 5 models (ChimeraX will only open the best scoring one by default) and see if it puts the peptide in the same or different locations. The 5 models are the .pdb files in directory ~/Downloads/ChimeraX/AlphaFold/prediction_N.
Alphafold cannot use the known structure of your protein, so each prediction has to rerun the full protein and peptide if you want to try different peptides.
As Elaine said looking for docking software that can handle peptide ligands may be a better approach. But it is easy to try the AlphaFold approach and see if it produces anything useful.
Tom
On Dec 20, 2023, at 10:06 AM, Parks, Ryssa via ChimeraX-users chimerax-users@cgl.ucsf.edu wrote:
Hello, I work in a lab that researches biopesticides in the form of insect neuropeptides, which bind to receptor proteins. I am trying to use ChimeraX to model potential binding sites between peptide ligands and their corresponding receptor. Does ChimeraX have this capability? When I was trying to figure this out myself online through the ChimeraX Youtube tutorials and tutorials on the UCSF site, I came across a comment on a video that mentioned I could upload the sequences of my receptor and peptide ligand separated by a comma, and model it as a protein complex, and then I can get information on where the two molecules interact. I’m relatively new to protein research, and so I hadn’t yet thought of ligand-protein interactions as a type of protein complex. My specific questions are:
· When modeling a peptide ligand and a receptor as a protein complex, will ChimeraX accurately predict the relationship between the two molecules based on binding compatibility? Or is it based on something else, like similarity of sequence alignments…? · If multiple binding sites are possible, is there any way to view the different options for multiple binding sites? Or is there any way to restrict the area on the receptor to which my target peptide ligand is predicted to bind (i.e. the extracellular region only)? · The peptide ligands are only 7-9 amino acids in length and do not take very long to model, but the receptor is about 600 amino acids and takes a much longer time to model. Do I have to model protein complexes by re-modeling everything for each different receptor-ligand combination, or can I combine existing models somehow?
I hope these questions aren’t too troublesome, and I hope I haven’t grossly misunderstood ChimeraX’s capabilities.
Thank you for your time,
Ryssa Parks
ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu mailto:chimerax-users@cgl.ucsf.edu To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu mailto:chimerax-users-leave@cgl.ucsf.edu Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/
participants (3)
-
Elaine Meng -
Parks, Ryssa -
Tom Goddard