Hi all, recently I have done ad-initio process of homologous protein of GroEL. But the result structure seems that the topview merges with sideview.
For example, the sideview of GroEL should be oblong shape, having a typical four sticks-like feature, and the topview should be circular. But the result shows a cicular sideview with four sticks-like feature, and the topview are also more similar to oblong shape than round shape.
Does anyone have suggestions about this issue? I don’t know whether this issue is caused by preferred-orientation (tilt-view is not enough) or unsuitable parameters (I used default parameters)?
Any help will be appreciated. Thanks.
Hi,
I have found some similar problems with one dataset, and it got slightly better by setting the class similarity to a higher value (0.9 or higher)!
Hi Rafael-Ayala and Morlyate,
It’s interesting that the presence of a preferred orientation causes cryoSPARC problems. Remember that the top view is a single view whereas there are actually many different side views. All of the top view images contribute to a single slice in Fourier space while the side views do all of the real work and fill up the 3D Fourier transform (the side views being similar to a single axis tomogram). In working with images of 20S proteasomes, that have the same problem (top views and side views) I found that I could remove all of the top views after 2D classification without hurting the resolution of the 3D map. This observation isn’t surprising because, as I mentioned, the top views aren’t contributing much information. For the 20S it wasn’t necessary but it could be that 20S side views and top views are more different than GroEL side views and top views.
Oh, thanks a lot. I will have a try.
Very impressive! I haven’t known that before. Maybe I can also have this try. Thanks a lot!