Hi Tom. thanks for the clear explanation. So, it seems that the intelligence behind Boltz is very similar to that of a human and capricious... Thanks again, Marco Il 05/12/2025 19:33, Tom Goddard ha scritto:
Hi Marco,
Boltz allows specifying where a ligand should bind. It is called a "pocket constraint" where you can specify residue numbers in the protein that the ligand should be near. ChimeraX doesn't have a user interface to set this. But you could still run a ChimeraX prediction without a pocket constraint, then edit the Boltz .yaml input text file produced by ChimeraX to add the pocket constraint, and rerun Boltz from a terminal. I described how to do a similar rerunning to add modified protein residues in this mailing list message
https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/... <https://urldefense.com/v3/__https://mail.cgl.ucsf.edu/mailman/archives/list/...>
The Boltz documentation for specifying a pocket constraint is here
https://github.com/jwohlwend/boltz/blob/main/docs/prediction.md <https://urldefense.com/v3/__https://github.com/jwohlwend/boltz/blob/main/doc...>
I plan to add ChimeraX user interfaces to do these fancier Boltz predictions such as pocket restraints, atom-atom distance restraints, covalently bound ligands, and post-translational modifications, but I haven't done it yet. There is a reason I have been a bit slow to add those features. When I have used those capabilities when running Boltz by hand they very often gave bad results. For example, I had a ligand binding prediction that put the ligand in the wrong place. So I added a pocket constraint and then it did put the ligand in the right pocket. But the pose and the binding affinity were all wrong -- I knew because I had an experimental structure. My suspicion is that if you have to tell Boltz to put the ligand somewhere where it doesn't want to put it, the chance of getting a good pose or affinity prediction is small. There is a reason why it didn't put the ligand there without the constraint -- it doesn't think the ligand goes there, and forcing it only has a small chance of allowing Boltz to figure out the right binding pose.
Tom
On Dec 5, 2025, at 1:30 AM, Marco Sette via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi all,
does Boltz allows to define the binding region?
Thanks for your help,
Best regards,
Marco Sette
-- Marco Sette, PhD Associate Professor of Molecular Biology Department of Chemical Sciences and Technology The University of Rome, "Tor Vergata" Via della Ricerca Scientifica, 00133, Rome, Italy e-mail: sette@uniroma2.it Tel.: +39-0672594424 Fax: +39-0672594328
_______________________________________________ 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/ <https://urldefense.com/v3/__https://mail.cgl.ucsf.edu/mailman/archives/list/...>
-- Marco Sette, PhD Associate Professor of Molecular Biology Department of Chemical Sciences and Technology The University of Rome, "Tor Vergata" Via della Ricerca Scientifica, 00133, Rome, Italy e-mail:sette@uniroma2.it Tel.: +39-0672594424 Fax: +39-0672594328 -- Questa email è stata esaminata alla ricerca di virus dal software antivirus Avast. www.avast.com