Hello everyone,
My name is Leo, an SPA beginner reaching out to you for some tips for processing my quite tricky sample.
Some info about it:
It is a small heterodimer of proteins A + B with a mass of ~80 kDa in total (~90 A at the longest axis). Protein A alone can form trimers, but dimers are preferred when B is present (suggested by SEC, SEC-MALS and ELISA, but at lower concentrations than used for grids). However, trimers were detected at higher concentrations used for SPA, which can be seen in 2D classification (red square selections in the 2D classes below).
I used an 8 mg/ml concentration and Cu300 R1.2/1.3 grids, cause the structure for A (trimer) alone was solved using a 10 mg/ml sample.
Unfortunately, it seems to really suffer from the AWI problem because it aggregates a lot without detergent and moves to the carbon as a huge aggregated net. Because of that, I used 0.15 nM DDM (see example micrographs below). There is a better distribution with DDM, but some aggregation is still present, and the protein still tends to avoid the thin parts of the ice, so the ice is quite thick in my data.
Data collection parameters:
Data (~18000 micrographs) was collected on Titan Krios 1 with K3, 0.836 A/pix, Energy Filter 20 eV, Total dose 50 (1 e/A2 per frame), Defocus -0.8 to -2.0.
Processing:
I attached some of the results I got after testing a 2000 micrograph batch from the full dataset that may be useful to get a better image of my data.
Particles were picked from the denoised micrographs with both the Blob Picker and Template Picker, using a 20A map generated from the AF model to prepare 50 templates. I am also planning to give Topaz a try.
For now, 2D classes I got do not look the worst (I guess), but I am having trouble getting any reasonable 3D volumes out of them, and separating trimers from the dimers even after a few rounds of 2D classification.
I would greatly appreciate any help, tips and comments from you. I will be happy to provide more info if needed.
Thank you in advance!
Denoised mics and picking inspection:
300 px box first 2D classification (red - trimers, yellow - dimers):
Parameters:
-
Number of 2D classes: 200
-
Maximum resolution (A): 3
-
Initial classification uncertainty factor: 1
-
Circular mask diameter (A): 120
-
Number of final full iterations: 20
-
Number of online-EM iterations: 80
-
Batchsize per class: 400
-
Use clamp-solvent to solve 2D classes: true
300 px box (after a few rounds of 2D classification cleaning):
Parameters:
-
Number of 2D classes: 100
-
Maximum resolution (A): 4
-
Maximum alignment res (A): 6
-
Circular mask diameter (A): 120
-
Circular mask diameter outer (A): 150
-
Batchsize per class: 400
-
Number of online-EM iterations: 90
-
Number of final full iterations: 20
200 px box (after a few rounds of 2D classification cleaning):
Parameters:
-
Number of 2D classes: 100
-
Maximum resolution (A): 3
-
Circular mask diameter (A): 120
-
Circular mask diameter outer (A): 150
-
Number of online-EM iterations: 80
-
Number of final full iterations: 20
-
Batchsize per class: 200
-
Force max over poses/shifts: false
Ab-initio (using 2D classes from 300 px particle set):
6 Custom Parameters:
-
Number of Ab-Initio classes: 5
-
Maximum resolution (Angstroms): 10
-
Window diameter inner (A): 100
-
Final minibatch size: 1000
-
Initial minibatch size: 300
-
Window diameter outer (A): 120
