Hi Guillaume,_______________________________________________I have been testing Boltz 2 with ligand binding affinity prediction in ChimeraX for the past week. I was about to put it in the daily build but yestereday Boltz 2.2.0 was released. So I am thinking about whether I release Boltz 2.1.1 which I have tested or spend this week trying out 2.2.0. It is hard to keep up with the Boltz developers. The Boltz developers fix lots of problems. So my ChimeraX Boltz 2 version differs only in small ways from theirs, adding a few extra options to better control Boltz. You can see my changes to my github fork of Boltz 2 hereLike most machine learning developers the Boltz team has expensive ($10,000 - $30,000) Nvidia AI graphics and use Linux so that tends to be the only configuration they think about. So I have to modify their code in small ways to get to install and run on Mac, Windows, and CPUs without Nvidia graphics. Another side-effect of the developers use of AI GPUs that few biologists have is that Boltz takes a lot of memory. If you have an Nvidia GPU with only 8 GB of memory I'm not sure you can do any prediction on Linux (although surprisingly on Windows you can because it gracefully fallsback to using CPU memory). On a Mac with 64 GB you can predict reasonable size structures (1000 residues and ligand atoms) at reasonable speeds.TomOn Jul 15, 2025, at 4:47 AM, Guillaume Gaullier <guillaume.gaullier@kemi.uu.se> wrote:Hi Tom,
This is great, I am excited to learn that this is eventually coming to ChimeraX!
Are the developers of Boltz at least willing to take your contributions to fix these compatibility problems? Otherwise it seems like quite a burden to have to maintain your set of fixes while their code will keep evolving.
Cheers,
Guillaume
On 1 Jul 2025, at 03:30, Tom Goddard via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi Saverio,_______________________________________________
As Elaine mentioned I am working on having ChimeraX use Boltz 2.1.1 and will put that in the ChimeraX daily build when it is available and will announce it on the mailing list and on BlueSky. There are currently a bunch of problems with Boltz 2, for instance it won't install on Mac, and might not install on Windows because its installation file (pyproject.toml) says it requires cuequivariance CUDA modules that are only available on Linux. It is easy to fix that problem, but there are many other problems, errors running on CPU, errors with Mac GPU, prediction quality issues, changed options names (--no_potentials became --use_potentials with opposite default behavior, --no_trifast became --no_kernels). The Boltz developers are apparently not too interested in making it work on anything but Linux with Nvidia graphics. But my aim in ChimeraX is that Mac, Windows, and Linux and without Nvidia graphics should all be usable. I have run tests and been able to fix many of the problems. Also there is new ChimeraX user interface needed to allow use of the new Boltz 2 capabilities like predicting ligand affinity. I guess I will have Boltz 2 in ChimeraX daily builds in the latter half of July.
Here is a ChimeraX ticket that describes the many Boltz 2 problems and I will be adding comments as I find fixes
https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/18061
Tom
On Jun 30, 2025, at 5:49 AM, saverio via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Hi,
it is possible to install the latest version ( 2.1.1 ) of boltz
in ChimeraX 1.10 with the command boltz install?
Thanks.
Saverio
_______________________________________________
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/
VARNING: Klicka inte på länkar och öppna inte bilagor om du inte känner igen avsändaren och vet att innehållet är säkert.
CAUTION: Do not click on links or open attachments unless you recognise the sender and know the content is safe.
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/
När du har kontakt med oss på Uppsala universitet med e-post så innebär det att vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du läsa här: http://www.uu.se/om-uu/dataskydd-personuppgifter/
E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy
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/