Gabriele Corso
Gabriele Corso
Hi Rui, yes it is possible but not currently supported on the repo. We will likely eventually get around to implementing it but if someone wants to contribute this as...
You can have an ipTM score even across non-polymer entities. This is computed by constructing frames for atoms in the molecule. You can take a look at the AlphaFold3 supplementary...
Hi all, pocket and contact conditioning support is now merged in the public repo (new release coming soon). On top of it, you can now force the constraint to be...
Yes, pLDDT is (0,1) but we changed it in the B-factor column because many visualization software are used to that scale. Is there any particular reason to also have it...
Hi @rjrich, no unfortunately at the moment sharding a single prediction across GPUs is not yet supported and unfortunately 2300 residues might be too much for a 48GB GPU at...
I would say probably something between 40 and 50 GB
That makes sense, what would you think the behavior should be in the case of a multi-complex prediction run in which some but not all the complexes fail?
Thanks for pointing this out, the latest code should now work!
Hi, in v0.3.2 we pushed a few memory improvements. Could you test whether this issue still occurs?
Hi @xiaolinpan, unfortunately at the moment we do not support the predictions of structures with >> 2000 residues/tokens on regular GPUs, the structure in your examples has above 3200 residues