I have obtained a ~4.5Å resolution map, which has helped determine the general direction of the main chain. The next step is to improve the resolution further by changing the symmetry and setting the initial low-pass resolution to 10Å. Would setting the initial low-pass resolution to 10Å introduce artifacts into the structure?
It could permit overfitting.
I generally avoid filtering any higher than 20A, although for local refinements where the structure is already clear, I may go to 12A or even 8A for a difficult or very small sample… but only for local refinements.
1 Like
Is there a significant difference between using 12A for local refinement and NU-refinemetn? Does NU-refinement introduce noise?
NU Refine searches globally around the entire sampling sphere, while local refinement only searches within a defined range of already assigned angles for a particle projection.
If the particles contain a lot of junk, or alignment is difficult, I’ve found NU Refine can overfit, which often manifests as “tulip flower” effects around density.
Thanks a lot. your reply is very helpful.