Symmetry and ab initio refinement

Hi all,

I’m working on a medium-sized protein solubilized in a liponanoparticle. The protein contains a domain which we expect to sit in the membrane (but not span it), entering and exiting on the same leaflet, and a soluble domain. There is a crystal structure of the non-membrane domain, which forms a very distinct trimer of beta-helices containing C3 symmetry. Given the structural information available, is it appropriate to run ab initio refinement with C3 symmetry, or should I stick to C1?

I would try both, and if you get a sensible looking model in C3, run a homogeneous refinement in C1 afterwards to confirm the enforced symmetry is real.


Follow up to this,

I’m seeing good results using the following workflow:

  1. Ab initio (C3 symmetry, 6 classes to look for gross structural differences/bad particles)
  2. Homogenous/NU-refinment (C3 symmetry, particles from classes that look good and aren’t super different)
  3. Symmetry expansion
  4. Particle subtraction using hand-generated masks
  5. Local refinement (C1)
  6. Iterate back to 4 if new density pops up

I’m mostly just posting this as a sanity check and to make sure that there aren’t any issues doing particle subtraction from particle sets have been previously symmetry expanded.

Also, is it typical to have to iterate through steps 3-5, similar to how one would process crystallography data? For the non membrane portion of my protein I’m seeing some sidechains and some ligand - super exciting!

I also wanted to ask if it makes sense to mask a single monomer after symmetry refinement? Or each of the discrete domains from the trimer?

1 Like