I had a dataset collected with EPU using maximum fractions. I want to bin the fractions into fewer numbers. How should I do it? What impact will the binning have on motion correction?
Welcome. 
Did you collect in EER, or MRC, or TIFF?
If MRC or TIFF, you have no choice - it’s already set. You have no more choice. Just tell CryoSPARC the total dose set at acquisition time.
For EER, I would recommend either leaving it unchanged at 40 (fractionation can be changed later during Reference-based Motion Correction if necessary) or starting with the same number of EER fractions as your total electron dose (i.e. if 50 e-/A^2, set EER fractions to 50).
To answer your other question, yes, fractionation will have an impact on motion correction - you need enough signal in each frame for alignment to proceed smoothly. In practice, the signal required can vary quite a bit depending on your sample, but a good general starting point is a group of ~1e-/A^2/frame (as described in the above paragraph).
Thank you for your answer! I did collect in TIFF format. The EPU used 1e/px for their auto determination of fractions and frames; one of the datasets ended with only 17 fractions. I want to explore the appropriate binning for this case. Dose motion corr2 have a bin function?
I wouldn’t bin/group at all for TIFF files.
But yes, MotCor2 can bin/group.