Hi CryoSPARC community,
I have collected a data set with 11184 micrographs with pixel size 0.788A/pix (super-res 0.3940 A/pix) and total dose of 50 e/A2 and tilted 40 degree. At the end of 2D classifications it ends up with 41K particles while other data set for the same sample, same number of micrographs and same data collection parameters but un-tilted ends up with 410K high quality particles. The data processing parameters are the same for both data sets because they are exactly the same sample, grid type, sample prep condition and concentrations.
I would appreciate any thoughts and suggestions.
total number of picks? were particles separated/monodisperse in both cases? is it thick ice? are you being hyper-selective for “really good 2D class”?
tilted data has longer path-length of ice so the resolutions will be slightly less, but I would not expect it to impact 2D very extremely, just ~1Ă… worse final reconstruction.
my guess is there is a lot of good particles in the 2D unselected. try 2D of the unselected particles and see if good classes come out. you can always input particle subsets to “inspect picks” tool and see where they come from in the dataset
How was the data collected? Was image- or stage-shift used to navigate between nearby holes? If the former, how well were changes in sample height compensated for? Do the estimated defoci fall reasonably within a nominal range and are the defocus gradients across your micrographs consistent with a 40-deg stage tilt?
What looks uniform and flat when perpendicular to the beam may end up being anything but when tilted so much. There is significant potential for a lot of overlap. Had this recently myself with a trial-run sample; bad orientation preference, but needed to drop concentration to avoid particles overlapping when tilted.