Hi, there,
Here are some features about my data: 180K particles; ~4 angstrom map; cFAR, 0.42; SCF, 0.89. According to Job: Orientation Diagnostics | CryoSPARC Guide, it seems that I should discard more junk or contaminants. However, I have already run heterogenous refinements using three junk models and one good model for more than ten rounds that led to only ~5K junk particles left. Also, I have already tried local refinment using diffrent parameter settings, i.e. Gaussian prior or not, different rotation or shift search extents, which didn’t improve map quality with reslution around 4 angstroms at all. Can anyone give some hints for what I should do next, or should I go to modify protein samples for new round of data collection? I would appreciate your response.
Best,
Lau
did you try 3DVA/3D classification to see if there is significant conformational heterogeneity? does 2D classification of final clean particles with default settings (200classes) yield mostly well-classified, obvious particles?
Thank you, CryoEM2, for your response.
Yes, we did 3DVA/3D classification and found that conformational heterogeneity occurs to some domains. However, subsequent Non-uniform Refinement based on the 3DVA/3D classification results didn’t improve map quality.
You can try more intensive 2D classification changing the maximum resolution from 6A to 3A. Sometimes this catches some more junk particles, but also allows you pick the best particles. This may increase the resolution and the orientation diagnostic metrics of your final map. Another thing to try is heterogenous refinement but using two copies of your final map as input - sometimes this can separate out particles into a high resolution class and a low resolution class. Good luck!
2 Likes