Safe map sharpening

Hi all,

I am a new cryoSPARC user and am enjoying the software quite a bit. However, I have a question on how to optimise the map sharpening.

I get a map that is so far at medium resolution (7.5 A global based on 0.143 FSC) and after a local resolution job and colouring of the refined and auto-sharpened map I get colouring indicating local resolution variations between 3 and 15 A more or less. However the regions that are supposed to be at the highest resolution look very smooth and fused and playing with the threshold does not seem to reveal any high resolution features (it basically looks like a low resolution, ~15A all around map). Since this is an auto sharpened map straight out from the final homogeneous refinement job, I am wondering how to best perform the sharpening without introducing artefacts. I tried increasing the maximum resolution in the sharpening job to the best resolution picked by the local res estimation but that introduced great artifactual noise to the structure. I also tried lowering the B-factor further than the one from the last Guinier plot of the refinement job but I am not sure how far it is safe to go. I am probably missing something fundamental but any suggestions would be greatly appreciated.



Hello @PVK,

It is fairly common for parts of a map to be under/over-sharpened in cases that have high variance in local resolutions, as you described. This is because in homogenous refinement, a single B-factor is estimated for the entire structure. For cases like this, I would recommend trying a Non-uniform Refinement to improve the overall map resolution (as opposed to using a homogenous refinement). Then, the locally-filtered output of the Non-Uniform Refinement can reveal more detail in the higher-resolution parts of the structure. This algorithm is designed to account for large variances in resolution.

To improve the interpretability of your existing volume, I would recommend trying a wide range of B-factors and visually inspecting the resulting maps from each. Since the resolutions are very different (7.5A to 3A), the B-factors that work for each part of the structure will be different. At high sharpening values (more negative values), you won’t be able to interpret the low-resolution areas of the map but the high-resolution areas will appear better, and vice versa.

I hope that helps - please let me know if you have any more questions.


Dear Ali,

Thanks for the quick reply. We are installing the latest version of CS tomorrow and I will definitely try non-uniform refinement as well as local masking. Thanks again for the suggestion.