Wrong CTF estimation (CTFFIND4) after MotionCor2(BETA) with image binning of 2

request_recorded
open

#1

I tested the MotionCor2(BETA) wrapper in v2.5.0 with image binning factor of 2. it runs successfully, However, use the aligned micrographs as the input for the CTF estimation (CTFFIND4) job, apparently, the CTF estimates are wrong. my guess is the micrograph pixel size parameter for the CTF estimation job is inherited from Import Movie job, not the pixel size after binning. Can you add the option to manually input pixel size information for CTF estimation job?


#2

Hi @bing,

Thank you for pointing this out! We’ll make sure this bug is fixed immediately.


#3

At least in my hands, the dose estimation is also wrong. When importing movies, I specify the total dose, and the MotionCor2 wrapper uses it as input for the -FmDose parameter, which should be per frame dose.

Additionally, when running with TIF movies and an MRC gain reference, it fails with an error every time (sent via PM to Suhail)

Cheers
Oli


#4

Hi @sarulthasan,

Thanks for the prompt reply. Another highly expected change is that CryoSparc will handle TIFF format in a more conventional way, I think @apunjani talked about this in a post. The origin and direction of y-axis of TIFF and MRC are opposite, Other EM tools such as IMOD, EMAN, and MotionCor2 will flip the Y-axis whenever read or write the TIFF file.


#5

Hi,
Sorry if this is a basic question… I just started processing a dataset that is in tiff format with image binning of 2. when importing the movies do i imput the binned or unbinned pixel size? and also should i use the total dose or the dose per frame? reading above I am a bit confused…
thanks!


#6

Hi @BWise, if the TIFF files are already binned, you should input the binned pixel size (it’s just as if you had collected the data on a sensor with less pixels that are larger). The dose in cryoSPARC should be input as the total dose across all frames in the movie.


#7

ok, i will do that.
thanks a lot for your help :slightly_smiling_face: