I’m trying out local refinement on a subregion of my density for the first time, and I would appreciate some advice on what region to mask as well as the location of the fulcrum and some other parameters.
Our protein is a nonameric ring, where each subunit crosses over its neighbouring subunit at some point with a stretch of amino acids. Would it offer any benefit to mask over two subunits? or is one subunit the best option?
Since our protein is a ring, each subunit has two points of contact with the rest of the protein. What is the ideal/logical region for the fulcrum?
I initially though to keep the default, which would be the center of the protein, if I understand correctly, but now I’m not so sure about what the best option would be. Since the subunits are identical, I wonder if there is a fulcrum [and/or a rotation search extent and shift search extent], that I can choose that would make CryoSPARC use each and every subunit’s information in every particle image towards resolving one subunit. I imagine this would give me 9x the information from every particle image. I.e., Every subunit in a particle image would be analyzed and accepted for the refinement of the subunit I’ve included in the mask. I’m not sure if I’ve worded this clearly enough. I’m hoping this might perhaps result in a higher resolution.
My initial thoughts were to set the rotation search extent to 360 deg and the shift search extent to the inner diameter of the ring (~55A) but I feel like this might be too much. I’m not sure if it’s a reasonable choice or not and I fear that I may be misunderstanding the concept of search extents.
I also am thinking to avoid particle subtraction, since all the areas of my map’s electron density has information that is/could be useful
In addition, each monomer has 2 ligands bound to it. I was thinking that after this local refinement I could re-run it while setting the fulcrum between the ligands and it’s adjacent subunit to try and push the resolution of the ligand region a little further. Is this a reasonable next step?
I dont think this is how local refine works. you will want to perform symmetry expansion prior and then treat this as a single subunit. I think local refine works extremely poorly outside the range of 4-5 degrees and 4-5A translation - it’s not a robust platform to be completely moving and twisting entire proteins to align them. maybe I’m missing the crux of your workflow, I very infrequently work with symmetric complexes
I have a few maps ready, some of which have C9 symmetry imposed, and some without it. So I can try using the maps where I didn’t impose symmetry and I can use the output from a completed symmetry expansion job so that’s also ready - but do you mind clarifying what you mean by treating this as a single subunit? I haven’t worked with local refinement on a single subunit so I’m not sure what that entails, and that’s kind of what I’m trying to understand.
The guide and case study on CryoSPARC’s website discuss a situation where the area of interest is connected to the rest of the protein from only one side, so I’m not sure where the fulcrum or search extents should be for my case.
I don’t think you’re missing anything from my workflow, I’m missing the best way to deal with this case.
After extracting my particles I ran 2d classification, selected classes, ran ab initio, took the good class and ran it through NU-ref with and without symmetry imposed. I have some other workflows where I ran heterogeneous refinement and selected some volumes from there to select a subset of the confirmations, but this is all mostly the standard workflow I think.
once you have symmetry expanded, you will have 9x the number of particles, each will have been rotated along symmetry axis such that every “monomer” now has the data for all 9 positions in one place. you should be using a mask that only covers 1/9th of the protein. and you should be trying to refine all 9 copies of the individual particle for ALL particles to a single structure. Not sure of best fulcrum but I imagine center of mask and center of box are reasonable starting points. use 5 and 5 for shifts and rotations in a first pass but hopefully someone more experienced can chime in with relevant workflow and parameters