I know relion and cryosparc have different conventions for Dn symmetry and we need to transform the cs file with pyem before we import Dn particles from Cryosparc into Relion. But I wonder if we need to transform the star file before we import Dn particles from Relion into Cryosparc. It seems that the reconstruction in Cryosparc has no obvious errors even if no transformation is applied to the star file before I import them from relion into cryosparc. Does Cryosparc do the transformation automatically when importing Dn particles from Relion? Thank you for your answers.
Thanks for reporting. We confirmed that the Dn symmetry convention in CryoSPARC does have a 90° rotation around the z-axis compared to RELION for D symmetry. Specifically, RELION has the 2-fold axis on the x-axis, whereas CryoSPARC has it on the Y axis.
Because of this, it wouldn’t be expected that in general particles’ poses from RELION that were refined with Dn symmetry would adhere to CryoSPARC’s convention. If the imported poses are intended to be used downstream in CryoSPARC, it would generally be required to import both the particles and volume output from the RELION refinement, and to run Volume Alignment Tools on particles+volume with “Symmetry Alignment” for the Dn symmetry enabled.
One exception is that for Dn groups where n is a multiple of 4, the volume is symmetric under a 90º rotation around Z, so these groups coincidentally will be equivalent between RELION and CS.
Can you confirm which Dn group you observed the following behaviour for? If n isn’t a multiple of four, the lack of obvious problems is unexpected, and I’d want to know more information about the specific sequence of jobs you ran.