Weird 2D classification

Hello Cryosparc community. I am trying to get 2D classification of my particles and when I run the job i get blurry blobs that do not look like the particles we are expecting. Do you have any idea what could be happening here? Our particles are 450kDa and around 15nm diameter.

These are the parameters in the 2D classification job:

window inner radius: 0.85
window outer radius: 0.99
2D classes: 25 (I’ve tried with 100 2D and the result is the same)
max res 6A
max align none
initial uncert. factor: 2
circular mask diameter: none
circular mask diameter outer: none
recenter 2d classes: yes
recenter mask threshold: 0.2
force max over poses/shifts: yes
(alll next default).

What do the micrographs look like? Do you have clear particles, easily visually separable from background? And how does the picking look, do your picks look like they are picking most of the particles and not too many non particles? Do CTF fits look reasonable?

1 Like

We had a problem with micrograph hole shapes. They were not circular, they were oval instead. Thus, in almost all micrographs there is the carbon edge of the hole (15-20% of all the micrographs is the carbon edge of the hole). The autopicking always takes particles from there. Normal particles in the ice are clearly seen and picked too. Also CTF looks great.

Ah - if there is carbon in the frame, CTF will always look great - I would maybe try excluding all exposures with carbon as a starting point then, just to see if that is the issue, then proceed from there