K3 Super Resolution Scaling

bell489 · January 29, 2020, 8:19pm

Hi,

I am working with some data collected on a Krios microscope with a K3 camera. I can successfully import the collected movies, perform motion correction, and perform CTF estimation. However, when I go to manually pick particles, the micrographs are very scaled down and do not encompass the entire work space. I was still able to pick particles and make initial 2D class averages for automated particle picking, but automated particle picking was not able to successfully find the particles in the micrograph. In my previous data sets, collected on a Glacios, this was never an issue. Is there something I am missing with processing super resolution data from a K3 camera?

Thanks!

sdawood · January 29, 2020, 8:38pm

Hi @bell489,

Thanks for reporting this. In v2.14 (next release), we’ll be introducing an improved manual picker with the ability to zoom and pan micrographs. I’ll update this post when it has been released.

Thanks,
Suhail

bell489 · January 29, 2020, 9:19pm

Hi,

Thank you for your reply. Just to be sure, there is no way I can move forward with my processing until v2.14? Is there something specific you have to do when importing K3 data, such as changing specific parameters from their normal default settings?

Thanks!

DanielAsarnow · January 29, 2020, 9:48pm

I imagine the issue is just that the micrographs at super-resolution are huge, so the whole thing isn’t being displayed. Typically we down-scale by 2x to the physical pixel size during motion correction, since you don’t often see sub-Angstrom resolutions in single-particle EM

I also know that the template/blob pickers still go all over the micrographs, so maybe just use the part you can see to control the pick quality and try that instead.

sdawood · January 29, 2020, 11:00pm

Hi @bell489,

Sorry, I misunderstood your original issue; I thought you were able to see and manually pick particles via the interface, just not see the details due to the lack of zoom functionality. Could you reply to this post or email us at support@structura.bio with a screenshot of the web application in this state?

Thanks,
Suhail

DanielAsarnow · January 30, 2020, 12:04am

Maybe I misunderstood; maybe “workspace” means the browser window. Then it’s indeed the lack of zoom. As a short-term workaround you can still use browser scaling (Ctrl-+) to zoom in.

bell489 · January 30, 2020, 4:01pm

Hi,

Here is a screenshot of what I am seeing in the web application. The micrographs are down scaled, but I can still pick particles with a box size of ~800 pixels. I can use these manually picked particles to make initial 2D class averages for automated particle picking, but that job is unsuccessful at finding my particles.

Thanks.

sdawood · January 30, 2020, 5:30pm

Hi @bell489,

Ah, so you’re able to extract manual picks but are having trouble with template picking?

You can also try the ‘ring’ shape option in the blob picker (introduced in v2.13) for another approach to automated particle picking.

Regards,
Suhail

bell489 · January 30, 2020, 8:35pm

I will give the blob picker a try thank you!

mplockhartuom · February 4, 2020, 12:30pm

I’m not sure if this is the right thread for this but I’ve been having problems with K3 data since the update to 2.13.2. When I template pick or blob pick the images don’t appear correctly and the particles are not scaled to the images:

I can’t seem to get anywhere with the particle inspector either:

I’ve tried downgrading and re-updating but I don’t seem to be getting anywhere. Previously I have put K3 corrected images into cryoSPARC (2.13.0) and it was all fine:

That said, when I did update (and down grade) cryosparcm status now gives me lots of errors. I assume they are connected:

----------------------------------------------------------------------------
CryoSPARC System master node installed at
/home/cryosparc_user/cryosparc/cryosparc2_master
Current cryoSPARC version: v2.13.2
----------------------------------------------------------------------------

cryosparcm process status:

app                              STOPPED   Not started
app_dev                          STOPPED   Not started
command_core                     RUNNING   pid 35653, uptime 0:12:32
command_proxy                    RUNNING   pid 37072, uptime 0:12:27
command_rtp                      STOPPED   Not started
command_vis                      RUNNING   pid 37022, uptime 0:12:28
database                         RUNNING   pid 34502, uptime 0:12:35
watchdog_dev                     STOPPED   Not started
webapp                           RUNNING   pid 37115, uptime 0:12:26
webapp_dev                       STOPPED   Not started

----------------------------------------------------------------------------

global config variables:

export CRYOSPARC_LICENSE_ID="XXXX"
export CRYOSPARC_MASTER_HOSTNAME="XXXX"
export CRYOSPARC_DB_PATH="/home/cryosparc_user/cryosparc/cryosparc2_database"
export CRYOSPARC_BASE_PORT=39000
export CRYOSPARC_DEVELOP=false
export CRYOSPARC_INSECURE=false
export CRYOSPARC_CLICK_WRAP=true

Any help would be greatly appreciated.

Thanks
Mike

update
Just to say im working with Ubuntu 18.04.

DanielAsarnow · February 4, 2020, 5:53pm

What’s the size of the images?

apunjani · February 4, 2020, 6:56pm

Hi @mplockhartuom,

Can you let us know the size of the images (i.e. superres?) and the pixel size?
Did you import these movies newly in v2.13, or were they imported before the update?

The output of cryosparcm looks normal - the “STOPPED” processes are not supposed to be running a production instance of cryoSPARC.

mplockhartuom · February 4, 2020, 7:22pm

They were imaged as super res but binned to 0.83 A/pix during motion correction with motioncor2 and imported. I’ve re-run the picking on newly imported images and on images that were imported prior to the update but the results were the same.

apunjani · February 5, 2020, 7:55pm

@mplockhartuom would you be able to share some of the problem data with us (movies and gain references + micrographs after motioncor2) in order to debug this? We cannot reproduce on super-res K3 data here (with or without binning). I will email you with upload credentials!

apunjani · February 5, 2020, 10:26pm

Hi @mplockhartuom and others who may be having this issue:
on our side we can reproduce this for micrographs that are externally motion corrected then imported. But if you import the raw movies into cryoSPARC and perform patch-motion (with eg. f-crop factor of 1/2), patch-CTF and then picking, everything works fine and picks show as they should

apunjani · February 6, 2020, 4:27pm

This is now fixed and will be released in the next update.