On 7/25/23 6:52 AM, g.m.tuveri--- via modeller_usage wrote: > Hello, I'm trying to add some missing domains to a big homodimeric > protein, pdb 8em4. I predicted the missing structures with Alphafold > and now I want to create an entire model of 8em4 using multiple > templates. ... > I keep obtaining knots, especially in the long domain.
I suggest you look at the initial model (file with a .ini extension, which by default is a regular PDB file). This is filled in by copying coordinates from the templates, so if your templates are in different orientations (or you have large insertions, which will be constructed in extended chain conformation) you may have atoms on top of each other, and the optimizer will likely be unable to recover. If this is the case, try structurally aligning your templates beforehand or providing your own initial model (https://salilab.org/modeller/10.4/manual/node27.html)
Ben Webb, Modeller Caretaker