Many small holes on the 2D images

Hi, all. The data is colletted from “Titan krios & K3” and some parameters are shown as bellow.
menbrance protein: ~70KD
pixel size: 0.85
Total exposure dose (e/A^2): 70

Dose anyone konw why the 2D images look so weird?
Thank you very much for you help!
Wei

I’ve seen similar. What happens if you select a couple classes and do another round? My guess is there are too many particles in each class, the alignment is poor. Any one of these classes will probably yield a number of different views in a next round of classification. You can try to split the dataset or use more classes, see if it helps.

What is your box size for extraction? It looks like it may be too small

Oli

I saw similar thing for my 100 kD protein, i found suggestion from Cianfrocco lab is helpful. Especially the use clamp solvant
https://cianfroccolab.lsi.umich.edu/protocols/cryosparc
Qianqian

1 Like

Hi,

for a 70KDa membrane protein, it means that the detergent has a major impact on signal and thus alignment. I think it is partly what happened, in addition to the excellent advices given above.
I would increase the amount of classes to search for (77K particles in 1 class makes it difficult to see unless your protein is truly stable), increase the amount of EM iterations (50) and batch size per class (400) with Force Max over poses off.
It also looks like you might have small aggregates or close neighbors. Playing with box size of mask for centering might help too.
Good luck
Vincent

1 Like