Edit Project Pixel Size

[Jacob Croft]

Hello,

If you have data processed with a certain pixel size, and then you later determine a more accurately calibrated pixel size for the same data, I was wondering what the easiest way to update that metadata in a project to continue/reprocess the data is.  The calibrated pixel size is 0.14 A off from the inaccurate pixel size, 1.72 (calibrated pixel size) vs 1.86 A/pix (what I specified when I imported the data).

Is it necessary to redo tomogram reconstruction?  I assume it is important to re-run CTF estimation.  If I simply "Edit project" and put the new info in and then re-run CTF estimation, will this new pixel size info info be applied if I then re-extract a particle set and then continue from there?

Best,

Jake

[Muyuan Chen]

I think it is only necessary to change the pixel size of the final map. 

The defocus value estimated by the CTF program would change with an updated pixel size, but the fitting, and phase flipping operation should not be impacted, and the particles should be identical other than its header information. 

Probably need to change the reported resolution as well...

Muyuan

--
--
----------------------------------------------------------------------------------------------
You received this message because you are subscribed to the Google
Groups "EMAN2" group.
To post to this group, send email to em...@googlegroups.com
To unsubscribe from this group, send email to eman2+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/eman2

---
You received this message because you are subscribed to the Google Groups "EMAN2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eman2+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/eman2/51261bf7-d855-4556-a688-647996973c36n%40googlegroups.com.

[Jacob Croft]

Thanks Muyuan.  For updating the resolution, is this as simple as re-plotting the FSC using the ratio of pixel sizes as a scaling factor on the x axis values?

Jake

To view this discussion visit https://groups.google.com/d/msgid/eman2/23937F17-4522-4FED-A635-3E00B38BC591%40gmail.com.

[Muyuan Chen]

Yes. 

To view this discussion visit https://groups.google.com/d/msgid/eman2/CAOa2YHAZaxWWcupdmfdWxzpwA0vNkHYmhP4UwY1ecUNCKyS86Q%40mail.gmail.com.

[Steve Ludtke]

A caveat to that. That is, indeed, fairly typical practice in single particle analysis, where the mis-calibration is only 1-2%, and that may be ok. However, strictly speaking a uniform scaling will not produce a valid CTF curve, and you can only approximate a correct curve by adjusting the defocus. It's only likely to be an impact at very high resolutions, but there is a potential issue with it.

To view this discussion visit https://groups.google.com/d/msgid/eman2/23937F17-4522-4FED-A635-3E00B38BC591%40gmail.com.

[Jacob Croft]

Thank you for the input, Steve.  Do you know if there would be an obvious sign that this is causing problems, i.e. weird artefacts in the map or a strange looking FSC?

To view this discussion visit https://groups.google.com/d/msgid/eman2/43E81CD1-9388-4C2B-8B1C-C62E82014526%40gmail.com.

[Steve Ludtke]

It really depends on the specific parameters of the problem. The attached image (below) is a simulation of a 10% magnification error. ie - where the sampling was believed to be 1.0 Å/pix but was actually 1.1 Å/pix. At 1.1 Å/pix, the defocus was 1 um. The best fit to the CTF curve at 1.0 A/pix is .842 microns. Due to the non-linearity of CTF zero crossing locations with respect to defocus, you can match at low resolution but then the zero crossings deviate at high resolution. In this simulation it begins causing significant problems between 2.5 and 3 Å resolution. The symptom would likely be the inability to refine the structure reliably past this resolution, but otherwise the structure should be fine. Obviously if the error were only 2%, the effect wouldn't hit until much higher resolution. Note that this also depends on your fitting strategy. If you focus on matching at high resolution, you can do so at the cost of less severe (but still significant) deviations at lower resolution.

To view this discussion visit https://groups.google.com/d/msgid/eman2/CAOa2YHDt_QUB7p-gNVckA23QNCfrCwhsyvOCFeMFSfROaMKJdg%40mail.gmail.com.