Improving flexible regions beyond local refinement


I was wondering if anyone had any suggestions or tips as to how I could improve the density (circled below) in my reconstruction- this is after signal subtraction and local refinement.

Clearly I have something there, but it’s a region that’s known to be very flexible and I’m having a lot of trouble improving the density. Using the same particle stack in relion to do focused classification yields basically the same overall shape but seemingly worse results and certainly the program isn’t really able to do much classification.

Any tips or suggestions for improving flexible densities post-local refinement?

1 Like

Have you tried for your initial global refinement to limit the resolution used for alignment? This can help improve regions that are much poorer than the best local resolution, or at least reduce the level of noise.

1 Like

Interesting idea, I’ll give that a shot and report back.


The result: limiting resolution to 9 A unfortunately produced an identical map

Hi @jdobbs. Out of curiosity, did you limit resolution in a NU-refinement job to 9 only, or did you also try higher values ? Cheers

Hi @marino-j. I tried 7 A and 9 A NU-refine jobs (with particle subtraction + local refinement), but without seeing any changes I did not continue and try further resolution limits.

I would try even lower - say 15Å

I can give that a shot! Also: In case anyone has a similar problem, I’ve taken a different approach since I posted this where I did 3DVA cluster mode followed by heterogenous refinement and local refinement on the “motion-classified” particles stacks. This seems to give me a good degree of improvement (though I’m at the early stages of this strategy).

Hopefully a 15 A res. limit will help too!

The outcome: while I’ve changed and optimized a few different settings, the 15Å resolution limit for NU-refinement seems to have made a good deal of difference! Thanks @olibclarke for the suggestion