Recentering classes on extraction?

hi all

is there a way to recenter particles on extraction? I have a long tube-shaped protein with larger mass on the end (like a hammer). However often my averages end up with the handle of the hammer in the center, while I want the head of the hammer.

In relion it is possible to re-extract particles given an X/Y/Z shift, but is this possible in cryosparc? A handy tool would be where you can open up your 2D averages and click each avg to set a new position for extraction, then use that as input for an extraction. Maybe it’s excessive, I dunno.


1 Like

I think that would be a wonderful tool to be able to define the recenter of the class averages before re-extraction. :smiley:

this would be great!

just to be sure… is there absolutely any way to recenter your particles in cryosparc?

I did a refinement and used the output particles from that to re-extract from the micrographs (hoping it would carry over the X/Y shift and other metadata). But the newly extracted particles still look extremely off-center when I follow by a 2D run. I suppose it’s possible I just have tons of incorrectly assigned angles in my initial recon…

if it is NOT possible to recenter then this could be a fundamental problem, as signal near the edge of the box may experience aliasing (particularly if the box is not big enough).


It should recenter on extraction according to the offsets associated with the particles (e.g. from 2D classification or 3D refinement), but there is no way to recenter on an arbitrary X.Y to my knowledge (as you described in your initial post).

If it isn’t recentering according to the offsets from refinement I would consider that a bug.


@orangeboomerang @olibclarke
CryoSPARC currently does not recentre particles on extraction but we do plan to implement this in a future version.

1 Like

Really?? Well that is good to know I had been assuming that it did… that is unfortunate

Yeah this is very important indeed. There should be an option to recenter, probably true by default but we should be able to set it off.

Coming in the next version, thanks @olibclarke @alburse


Following the same thread. I had problems recenter/separate closely packed ring shipped particles and moved on to the next step of design, but a colleague suggested that instead of reentering particles based on center of mass, perhaps we can recenter the particles based on continuity of ring density? Thanks.


Wondering if this has been implemented in the current version?



Dear @simonbrown,

As of v3.0+, you can re-center particles upon extraction by passing them back through an extract from micrographs job. Re-centering will be done by default if the particles have been previously aligned to 2D or 3D references (there is a parameter called “Recenter using aligned shift” that is active by default). This will use the shifts determined by any 3D reconstruction/refinement, or if not present, from 2D classification. Note that there isn’t yet an option to manually set the desired center location to some arbitrary location on the 3D or 2D reference.



Has the arbitrary re-centering been implemented?

Note that there isn’t yet an option to manually set the desired center location to some arbitrary location on the 3D or 2D reference

Hi @August,

For particles aligned to a 3D volume, arbitrary re-centering has been implemented via the Volume alignment tools job; please see the documentation for how to do this recentering by connecting volume alignment tools to extract from micrographs. Arbitrary re-centering hasn’t been implemented for 2D classes.