I Hi, I am writing to seek your advice regarding an issue I’ve encountered during 3D refinement in cryoSPARC. I’ve observed that the Global Spatial Frequency Spectrum Correlation (GSFSC) curve drops very rapidly at higher frequencies, and the overall resolution is not as high as expected.
Here are the steps I’ve already tried:
I increased the extraction box size, but this did not improve the GSFSC behavior.
I performed further particle sorting and used a subset of ~30,000 particles for refinement, which yielded a resolution of 3.66 Å (based on FSC=0.143). However, the GSFSC curve still falls off sharply at higher spatial frequencies.
I would greatly appreciate your insights into what might be causing this discrepancy—particularly why the GSFSC declines so rapidly despite achieving a moderate resolution. Could this be due to structural flexibility, CTF overfitting, or alignment inaccuracies? Any suggestions on how to improve high-frequency consistency would be very helpful.
Can you explain what you mean by “overall resolution is not as high as expected”? It looks to me as if you may have just reached the resolution of your refinement, the FSC doesn’t look particularly surprising to me (in fact it looks pretty good).
The minimum resolution value on the x-axis is determined by your pixel size (2x pixel size = Nyquist = right side cutoff of x-axis), so in this case it appears to me that you just have a pixel size that is much smaller than necessary. You don’t have tons of particles, especially if you are in C1 symmetry, so that could be limiting your resolution, which you could interrogate using the ResLog job. But in general, I don’t see any huge red flags here with the refinement and my guess is that this is just the best resolution that data set processes to.
I’ll add, if your expected resolution is higher because the max CTF resolution of the micrographs is higher, then it is probably some combination of flexibility of your target and not having enough particles. It also may depend on how large your particle is – smaller particles are harder to align and process. The max CTF resolution tells you how good the overall micrographs are, but doesn’t correlate to a direct prediction of the resolution you will be able to process to (I think of it as the maximum approximate resolution you can process to, if everything else is perfect, but not the resolution that will be attained, because life and EM are both not perfect)
