Source of alignments for global 3D classification

Hi everybody,

I’m getting different results from global 3D classification when I use homogeneous vs. non-uniform refinement as input. I know in the 3D classification job tutorial it says that the source of alignments3D is an important consideration, but I still find this surprising. I consistently see no difference in classes from NU-refine but am finally seeing differences from homogeneous refinement as input. Has anyone else run into this?

My protein has a large, stable region with 2 more mobile regions around it. I wonder if what I’m seeing with 3D is from how NU-refinement weights different areas of the map in the adaptive marginalization method, thus changing how particles are aligned. Every analysis I’ve tried downstream of NU-refinement (without particle subtraction) seems to have the problem of the alignments in my stable region outweighing anything else.

Thanks for any input!

When you say heterogeneous vs non-uniform do you mean homogeneous vs NU?

To check if it is adaptive marginalization or NU-weighting, you can turn both these options off independently in NU-refine

Yes, sorry! Meant to say homogeneous and not heterogeneous. Let me edit my original post to reflect this.

1 Like