FSC curve and preferred orientation?

Troubleshooting
1

Hi all,

My 2D result is quite beautiful. After hetero/homo refinement, the FSC curve can’t drop to 0. I’ve tried to remove the duplicate particles and it didn’t really help. All suggestions are welcomed.

Also any comments on the particle distribution?

map
map1920×1038 135 KB

Thanks,
Shasha

2

Hi Shasha,

What is your pixel size, and are you binning (Fourier cropping) your particles? It appears you are limited by sampling, so if you unbin you may see better results.

Re the orientation distribution, there is clearly a degree of preferred orientation, but to judge whether it is significant just look at the map - do you see clear, interpretable density, or is it stretched/streaky?

Cheers
Oli

3

Hi Oli,

Thanks for your reply!
Images are collect at 0.7888Å/pix (super-res 0.394Å/pix). When extracted particle, I binned 2X to 1.576Å/pix. Will try without binning.

Here I attached a figure of the map. It is not bad?

Thanks,
Shasha

Picture1
Picture1436×816 111 KB

4

It’s hard to tell from this snapshot, but I think yes it is stretched a bit - is that an alpha helix on the left? It looks like it is kind of smeared into the surrounding density, which would be consistent with the kind of artefacts seen with preferred orientation. To quantify, you can calculate a 3D-FSC.

Cheers
Oli

5

Hi,
Could you please explain the reason behind not dropping at zero if particles are binned? I am also facing the same issue. Without binning it’s getting drop to zero.

I am looking forward to hearing back from you.

Thanks,
Regards,
Priyanka

6

Hi Priyanka,

Resolution is limited by sampling (how big your pixel size is) - commonly referred to as the Nyquist limit (2x the pixel size).

Therefore, if your binned pixel size is larger than half the unbinned resolution of your dataset, the FSC for your binned data will not drop to zero at the Nyquist limit. This isn’t really a problem - you can always re-extract without binning if so desired.

(@rposert, maybe worth adding a paragraph to the guide in the section re FSC plots not dropping to zero to describe the case of resolution limited by sampling? Tutorial: Common CryoSPARC Plots - CryoSPARC Guide)

1 Like
7

Hi Oli,
Thanks for your quick reply and explanation. Even I am facing an issue of that the FSC could not even converge at final iteration. Is this problem can also be because of the same reason or mask (tight) can be the reason in this case?

I am looking forward to hearing back from you.
Thanks,
Priyanka

8

Not sure what you mean by the FSC not converging - perhaps you can post some images of the problematic FSC to help explain?

9

J481_fsc_iteration_003_after_fsc_mask_auto_tightenin_256_biiningg
Hi Oli,
This is how it looks if I binned the data.
I am looking to hear back from you.

Thanks,
Regards,
Priyanka

10

Yes, this is because your data is binned (downsampled), your resolution is limited by sampling as described above

2 Likes
11

Thanks for the suggestion @olibclarke — we’ve added a note to this effect to the guide page. I’ve also made a note to consider a more in-depth discussion of the FSC and resolution in a future update.

4 Likes