Refinement strategies for protein comprised of beta sheets


I have some issues refining the map of a protein (150kDa) that is mostly made up of beta sheets and has no symmetry (C1).
After multiple rounds of 2D and 3D classification to filter out bad particles, I have ~100k particles which I then used to create ab initio models. I used a maximum resolution of 5 angstroms and initial resolution of 12 angstroms, while keeping the values of the initial and final batchsize as 1000. Using these parameters I created a reasonable initial model.
Homogenous refinement using the default parameters as well as using a 6 - 12 angstrom initial lowpass resolution caused the map to be over-refined, producing ‘spiky’ maps without significant improvement to the protein features.
Non-uniformed refinement gave similar results. I also tried to provide a mask with soft padding width of 10 but there was no improvement.
I believe that the lack of alpha helices in the structure may be causing these problems due to the lack of any discerable features at medium resolution that affects the refinement process. Please correct me if I am wrong.

Any advice is appreciated!


1 Like

Very interested questions. I wonder if you have solved the problem? Thank you! Zhening