Hi Sam, It sounds to me like your Boltz prediction not running is because the Colabfold MSA server is not responding. Here is an issue on the Boltz github site from 12 hours ago reporting this problem https://github.com/jwohlwend/boltz/issues/629 When you run a Boltz prediction from ChimeraX it shows in the Boltz panel the progress it is making. For instance it will say "Prediction runing 14 seconds: sequence search submitted" "Prediction running 16 seconds: sequence search running" "Prediction running 22 seconds: structure inference" ... So if it is not finishing you should easily be able to see what step it is stuck on. My guess is you don't get past "sequence search submitted". But I just ran two Boltz predictions and the sequence searches completed in under a minute. So maybe the Colabfold server is back to working. We do not run that server. Your Windows installation problem where it does not create the mols folder is something I have not seen and is not related to the Colabfold server. You should try the install, and then use ChimeraX menu entry Help / Report a Bug... so I can see the exact error message. That step is Boltz downloading the PDB chemical component dictionary (3 and 5 letter codes for small molecules) as a .tar file from https://huggingface.co/boltz-community/boltz-2/resolve/main/mols.tar and then extracting the tar file. I don't know why it would fail without more info. Tom
On Dec 10, 2025, at 1:53 AM, Sam Allen via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hello Tom and co.,
I am currently trying to use ChimeraX Boltz tool to predict the structure of some small ligands onto a GPCR and my prediction running timer just never stops. I initially attempted to install and run the tool on Windows but the install failed to import the mols folder properly multiple times (uninstall and reinstalled), after manually untaring the folder in Powershell it still didn’t run (Timeout while submitting to MSA server). I then decided to switch over to my mac (M1 16GB) and while the install of Boltz in ChimeraX no longer had any error and the prediction didn’t error either, the timer just continued to go up (to 26000 eventually). I figured the protein was too large (471 residues, makes sense), so I tried to rerun one of your test examples (8rf4, https://www.rbvi.ucsf.edu/chimerax/data/boltz-apr2025/runtimes.html) and still no prediction output after 20 minutes. I’ve tried switching from default compute device to GPU or CPU and still no difference. Additionally, in activity monitor it shows ChimeraX is taking up nearly no GPU or CPU usage, so I’m confused as to what is happening. Do you have any advice or know anything that might be going wrong that I can fix?
Regards, Sam _______________________________________________ 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/