Heterogenous refinement

Hello,
I have 110 kDa protein along with RNA. I want to perform heterogeneous refinement, I have extracted particle with 320 Box size, So what range of refinement box size is recommended? Because by default it is 128 but I am not sure about it. Pixel size is 0.85 A, 105 K magnification and bining 2.
Next question is my homogenous refinement stop after 3,46 and curves are bit different. I have attached the image of the curves. It will be great if I get guidance for both the questions.

Thank you

Hello @nikhil,

This depends on the kind of heterogeneity you are trying to resolve. If you are trying to eliminate junk particles, then calculate the voxel size for ~20 angstroms max resolution or so (max resolution = 2X apix). If you are trying to separate different conformations, this can be set to ~12 or 8 angstroms. There is no use in including high resolution information at this point, it can be even deleterious - at least IMO.

1 Like

Because by default it is 128 but I am not sure about it. Pixel size is 0.85 A, 105 K magnification and bining 2.
Next question is my homogenous refinement stop after 3,46 and curves are bit different. I have attached the image of the curves. It will be great if I get guidance for both the questions.

The GSFSC from homogeneous refinement could be approaching Nyquist or the box size is cutting stuff out. ~110kDa, 320 pix, 0.85A | 320pix*0.85A/pix = 272A, 272A should be ~2x the size of your max diameter so maybe the box size is fine. Could experiment.

I do like multi-class ab-initio followed by heterogeneous refinement, but never increased the down sampling to 128pix for hetero - just because it takes so long. Most 3D classification jobs will work best without the high frequency information in the 6-20A range.

1 Like

Thank you will try according to this.

Thank you.
This homogenous from initial ab-initio where I have applied Fourier crop (360 Box size, 180 Fourier crop size)
I have run the 2D classes without Fourier crop. Will do ab-initio and homogeneous refinement respectively. Hopefully it should not approach Nyquist.

1 Like

Good luck @nikhil, the FSCs getting close to the right side was the only obvious thing. Hope it improves things, the RNA you have should be helpful for contrast.

1 Like