Final Volume Pixel Size Mismatch

Troubleshooting
1

Hi,
I processed a dataset using the following parameters:
Pixel size = 0.57 Å/pixel
Accelerating voltage = 300 kV
Spherical aberration = 2.7 mm
Total dose = 50 e⁻/Ų
Box size = 256 pixels (default used during particle extraction)
Estimated protein size = ~100 Å in diameter
After processing, I obtained the final volume:
cryosparc_P616_J41_class_01_final_volume.mrc

When I open this file in ChimeraX, it shows:
Grid size: 64×64×64
Pixel size: 2.28 Å/pixel
Displayed at level 0.591, step 1, values float32

However, when I try to align this map with an AlphaFold3-predicted model, the volume appears ‘too small’ , and maybe I need to scale it up by ~4× in ChimeraX to get a reasonable fit. This makes me suspect that the pixel size metadata in the MRC file might be incorrect.

Any advice or insights would be greatly appreciated!

Thanks in advance,

Yan

2

Welcome to the forum. :slight_smile:

That sounds like a heterogeneous refinement or 3D classification output, which are binned down to a specified box size. If you carry out a homogeneous reconstruction or refinement of that map and particle stack, does it output a 256 pixel box volume?

Also, at 0.57 Ang, 256 pixels is quite tight for a 100 Ang protein, with the box being <1.5x particle diameter…

3

Thank you for your helpful suggestions!

Just to provide a bit more context: I originally started with Ab-Initio Reconstruction, and already at that stage, the resulting volume appeared too small in ChimeraX. I then proceeded with Homogeneous Refinement using the same particle set, and obtained a final map:
cryosparc_P616_J267_005_volume_map_sharp.mrc

Grid size: 256×256×256
Pixel size: 0.57 Å/pixel

Even though this volume has the expected pixel size (0.57), it still appears visually too smallin ChimeraX when compared to the AlphaFold3-predicted model.

I just realized that in the Patch Motion Correction, I had set the F-crop factor = 1/2, could this explain the mismatch in physical size I’m seeing in ChimeraX?

4

It might do. I don’t use Fourier cropping during Motion Correction myself. But CryoSPARC is usually pretty good at automatically scaling pixel calibrations when cropping. What format is your data?

5

Just to clarify my data setup: I have three datasets in total , and one of them uses .eer format movies , while the other two are in .mrc format. Despite the difference in input formats, all three datasets show the same issue — the final volumes appear too small in ChimeraX.

6

Then unless you typo’d during Import setup, I’m at a loss, as it’s not an issue I’ve seen before on that magnitude. :frowning:

7

Thank you very much for your help!

In fact, I’ve previously processed another dataset with a similar workflow, but with a pixel size of 0.92 Å/pixel, and I didn’t encounter such an issue back then.

I’ll go back and carefully trace each step of the processing pipeline to see if I can identify where things might have gone wrong. If I find a solution or figure out the cause, I’ll definitely post it here for your and others’ reference.

1 Like