This is something new I encountered recently in “Helical Refine”. 2D averages of this filament look fine. Contaminants in this dataset can reach 3.5Å easily using default parameters, so the dataset quality is okay.
But for this thin filament (40-60Å in diameter), using the default parameter without “Non-uniform refinement”, I see this strange artifact on the outer region. While looking at 2D averages, this is not the case. It seems this artifact even dominates the particle alignment.
The helical symmetry is pretty simple: there is only one possibility and no question in this part. I also tried common pitfalls, such as a) setting GSFSC to a higher resolution; b) using a bit narrower masking; c) turning off non-uniform refinement. But none works. Any tips on how to get rid of this artifact?
The method used to initialize the helical refinement’s volume:
No volume input. I normally used the initial volume generated from cryoSPARC. It works better than a cylinder in most cases.
The claimed GSFSC resolution that the refinement achieves at termination (5-6Å ish) A plot of the FSC curves for the refinement
The line from the event log reading Refining with helical symmetrization degree of ___?
Refining with helical symmetry order of 14 enforced.
I did the calculation = box-size / helical rise
Hi, Recently I experience this problem too with couple of datasets. The common things in all of them is they are thin filament (40-60 Å diameter). The artifactual density becomes prominent at later iterations (see below). I have tried several things: tight masking, Override outer filament diameter for search, non-uniform refinement ON/OFF. The masking helped in one dataset, but none helped for others. Any hint would be appreciated? Thanks,
Hi @mmclean, Thanks for the reply. I’ve tested the refinement without masking as suggested for the case for which the helical parameters are known. It helped partially. I cloud get rid of the artifactual density closer to the filament, but resultant map now has such artifactual density away from the filament density. (Artifactual density seems symmetrical as the imposed helical symmetry). The problem with no masking is that the refinement doesn’t want to converge to high resolution map. I wonder if the ‘no convergence of refinement’ and the artifactual density are related?
Hello,
I have another related situation. Basically, I have a hollow tube which has density from ~80-100 Å but is entirely hollow from 0-80 Å. In Helical refinement job input parameters I accidentally set the outer diameter limit to 70 Å. What was returned to me in a final volume was a “flowery filament” shown by the gray density. I later on solved the correct structure (blue) using an outer filament diameter of 110 Å. Note how the gray “noise flower” filament is in the center and must be made up of entirely noise.
I just re-ran the job with the final symmetry of the correct blue filament in the image above. I kept the outer filament diameter as 70 Å. The result is pretty much the same but now the flower density is around the impossible 70 Å diameter density as well as in surrounding it closer to the actual diameter.
