Resolution become worse suddenly in local refinement

Local Refinement
1

Greeting there,

I am doing local refinement now with particle subtraction, static mask, mask center as fulcrum location, reset and minimize per-particle scale, the resolution become better and better until for example Iteration 5, but become worse suddenly from Iteration 6 (even worse than the Iteration 0).
Can anyone point me out the possible reason?

Thanks in advance,

2

Hi Yanhe,

In order to diagnose the problem, it would be helpful to see the FSC curves from the iteration before and after the issue occurred.

Cheers
Oli

3

Dear Oli,

P18_J929_fsc_iteration_004
Here is the FSC right before it suddenly become worse, I don’t have the FSC of worse Iteration.
In the overview, it said “Using Filter Radius 130.742 (11.554A) | Previous: 156.476 (9.654A)”,
so I just kill the job.
BTW, I am doing local refinement on subregion of my complex.

Thanks a lot,

4

Hi Yanhe,

Without seeing the FSC afterwards, it is difficult to say very much about what might be wrong.

However, the spherical and no-mask resolutions of 40-60Å are very suspicious - are you sure you are masking the right region? What was the resolution of the preceding refinement, and what initial lowpass resolution are you using in local refinement?

Cheers
Oli

5

Dear Oli,

I will repeat it and have it post when I done.

I thought the 40-60Å is because particle subtraction, right? I subtract signal from all the left area except for my interested subregion.
The resolution of the preceding refinement is around 11Å, the lowpass resolution I used in local refinement is 12Å.

Thanks a lot,

6

Hi Yanhe,

Local refinement in cryoSPARC is unlikely to work well at such low resolution, unless you are talking about an especially large complex with a correspondingly large subregion.

Have you tried local refinement without signal subtraction? I would always advise trying this before trying with signal subtraction.

7

Dear Oli,

You are totally correct, I am working on large complex with a correspondingly subregion.

When I do local refinement on another subregion, the signal subtraction give better result (7.6Å) when compare with no subtraction (9.6Å).
But I will try refinement with no subtraction and repeat the job running mentioned above. Then post here to have your advice.

Thank you so much,