ChimeraX / Alphafold docking question: Heterodimer and ligand
Hello I'm currently evaluating ChimeraX for non-commercial use. I'm interested in predicting a complex of an unknown ligand to a heterodimer (receptor) with a known structure, pdb file available. The help files mention: *For predicting a complex (multimer), the sequences of all chains in the complex must be given. The same sequence must be given multiple times if it occurs in multiple copies in the complex. The sequences can be specified either collectively as a model number chosen from the menu of currently open models (e.g. when that model contains multiple chains)* If I download the structure of the heterodimer first, does this mean I can then select this then add the primary sequence of the ligand to try to dock? If not, can I add all the primary sequences and tick 'use PDB templates when predicting structures'? Would this then use the known experimental structure of the receptor? I hope this makes sense and thanks for the help Best wishes Tony Anthony Newcombe Ballinascarthy Clonakilty Cork Ireland
Hi Tony, I wouldn't say that alphafold does docking per se; it will just predict the structure of the whole complex of protein chains that you give it. Furthermore, alphafold producing a complex doesn't mean that alphafold actually predicts that these proteins chains will bind one another; it is simply generating a complex because you told it to, and you'd have to look at the confidence metrics pLDDT (per-residue) and PAE (per residue-residue pair, particularly for the interchain pairs) to figure out whether you want to believe the results or not. There is some discussion/examples of how you might use Alphafold via ChimeraX to help figure out which proteins might bind one another here: <https://www.rbvi.ucsf.edu/chimerax/data/afbatch-jan2024/rim_dimers.html> That page discusses dimers only. In your case, you have a heterodimer + a third protein chain, so you'd need three sequences separated by commas in the Alphafold-Predict sequence entry field. If this is a large number of residues total, the calculation may take a very long time or may not even be feasible with a free Google account. I'm assuming your ligand is a protein or peptide. Otherwise, this is not possible with the current interface (i.e. it does not include nonprotein molecules). Even if you choose to use templates, it is not going to use the experimental structure wholesale, but merely use known structures as examples. I'm told that in practice, this makes suprisingly little difference, but you could certainly try both ways and see what you get in each case. If you really want to use a particular known experimental structure and not allow that part to change, one possibility would be to predict the other unknown structure (e.g. with Alphafold) and then use some dedicated protein-protein docking software on experimentally determined structure + the prediction. I don't have a good feel for how well that would work, however, and surely it depends on the system, amount of induced fit, etc. I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 8, 2024, at 1:02 PM, Anthony Newcombe via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello
I'm currently evaluating ChimeraX for non-commercial use. I'm interested in predicting a complex of an unknown ligand to a heterodimer (receptor) with a known structure, pdb file available. The help files mention:
For predicting a complex (multimer), the sequences of all chains in the complex must be given. The same sequence must be given multiple times if it occurs in multiple copies in the complex. The sequences can be specified either collectively as a model number chosen from the menu of currently open models (e.g. when that model contains multiple chains)
If I download the structure of the heterodimer first, does this mean I can then select this then add the primary sequence of the ligand to try to dock? If not, can I add all the primary sequences and tick 'use PDB templates when predicting structures'? Would this then use the known experimental structure of the receptor?
I hope this makes sense and thanks for the help
Best wishes
Tony
Anthony Newcombe Ballinascarthy Clonakilty Cork Ireland _______________________________________________ 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 Elaine Many thanks for the reply, this is most helpful Best wishes Tony On Wed, 8 May 2024, 9:58 pm Elaine Meng, <meng@cgl.ucsf.edu> wrote:
Hi Tony, I wouldn't say that alphafold does docking per se; it will just predict the structure of the whole complex of protein chains that you give it. Furthermore, alphafold producing a complex doesn't mean that alphafold actually predicts that these proteins chains will bind one another; it is simply generating a complex because you told it to, and you'd have to look at the confidence metrics pLDDT (per-residue) and PAE (per residue-residue pair, particularly for the interchain pairs) to figure out whether you want to believe the results or not.
There is some discussion/examples of how you might use Alphafold via ChimeraX to help figure out which proteins might bind one another here: <https://www.rbvi.ucsf.edu/chimerax/data/afbatch-jan2024/rim_dimers.html>
That page discusses dimers only. In your case, you have a heterodimer + a third protein chain, so you'd need three sequences separated by commas in the Alphafold-Predict sequence entry field. If this is a large number of residues total, the calculation may take a very long time or may not even be feasible with a free Google account. I'm assuming your ligand is a protein or peptide. Otherwise, this is not possible with the current interface (i.e. it does not include nonprotein molecules).
Even if you choose to use templates, it is not going to use the experimental structure wholesale, but merely use known structures as examples. I'm told that in practice, this makes suprisingly little difference, but you could certainly try both ways and see what you get in each case.
If you really want to use a particular known experimental structure and not allow that part to change, one possibility would be to predict the other unknown structure (e.g. with Alphafold) and then use some dedicated protein-protein docking software on experimentally determined structure + the prediction. I don't have a good feel for how well that would work, however, and surely it depends on the system, amount of induced fit, etc.
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On May 8, 2024, at 1:02 PM, Anthony Newcombe via ChimeraX-users < chimerax-users@cgl.ucsf.edu> wrote:
Hello
I'm currently evaluating ChimeraX for non-commercial use. I'm interested in predicting a complex of an unknown ligand to a heterodimer (receptor) with a known structure, pdb file available. The help files mention:
For predicting a complex (multimer), the sequences of all chains in the complex must be given. The same sequence must be given multiple times if it occurs in multiple copies in the complex. The sequences can be specified either collectively as a model number chosen from the menu of currently open models (e.g. when that model contains multiple chains)
If I download the structure of the heterodimer first, does this mean I can then select this then add the primary sequence of the ligand to try to dock? If not, can I add all the primary sequences and tick 'use PDB templates when predicting structures'? Would this then use the known experimental structure of the receptor?
I hope this makes sense and thanks for the help
Best wishes
Tony
Anthony Newcombe Ballinascarthy Clonakilty Cork Ireland _______________________________________________ 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/
participants (2)
-
Anthony Newcombe -
Elaine Meng