Error when own input mask


I have issues when trying to input my own mask on the Flex Mesh Prep. I have done the mask in two different ways and encounter different issues.
On the one hand, I have created the mask from NU-ref volumes by cropping and downsampling the volume to the parameters chosen for the 3D flex and I have gotten weird looking meshes as in the images below.

On the other side, I have created a mask from the Flex Data Prep consesus map and gotten the following error:
Traceback (most recent call last):
File “cryosparc_master/cryosparc_compute/”, line 96, in
File “cryosparc_master/cryosparc_compute/jobs/flex_refine/”, line 69, in
AssertionError: Input mask psize 2.0800 and input volume psize 2.0800 do not match

Has anyone experienced this issues? How do people usually input their own masks?


I haven’t run into the same issues exactly, but when I was first playing with 3DFlex I had a lot of trouble generating a mask. Ultimately I found it was a result of using Volume Tools to downsample the mask after creating it — I had much better success downsampling my initial (in your case NU refined) volume first, then creating the mask in the downsampled volume.

@are – I can’t comment on why the first approach seems to be providing those strange meshes, but I did want to note that this second AssertionError is a known issue in 3DFlex that should be fixed in an upcoming release. Unfortunately, we have an assertion that will throw an error if pixel sizes differ by minuscule amounts (up to machine precision) which can sometimes happen in certain workflows.

Thanks for the answer! In the end I am using the consensus map given on the first step of 3D flex to create the mask on ChimeraX using the eraser tool, as the issue with the pixel size comes from using the segmentation map and resampling.
Not sure if the weird looking first mesh has anything to do with that.

If I need to use a map from other jobs I will make sure I follow the steps on that order and hope I do not run in so many issues.