Leginon

I've noticed a problem like this if the bright and dark references are not up to date for the binning and dimension used in the image acquired to compute the dose in the presets manager (which is 512x512 with a binning of 1, I think?) and possibly also if bad pixels in the CCD are not marked in the Corrector node (also for the correct binning and dimension). Our camera has a big clump of bad pixels so this may not be so much of a problem for other labs. I think the problem is the consequence of having spikes in the image from bad pixels or pixels uncorrected for gain normalisation and they may change little or not at all for different exposure times,

William

"jriches" wrote: I've run into a couple of problems with the dose calibration and stigmator calibration on Leginon. The first problem is with the dose calibration. I've followed the instructions in the notes and I get a sensitivity of 3.3 counts per electron for our CCD camera, which seemed close enough to yours (4.6 from what I've written here) to be reasonable. However, when I go to the Presets Manager and press the button for determining the dose I get what seems like very large doses. This isn't bad in itself, but I don't seem to be able to reduce them by dropping the exposure time. When I drop the exposure time in half, the dose seems to drop by 10-20%, and likewise if I increase the exposure time by a factor of 3, the dose only increases by maybe 50%. I can probably muck around with spot size settings to get it about right, but I want to be able to trust the dose calibration in Leginon. The camera isn't anywhere near being saturated, so that isn't the cause of it.

The second problem regards the stigmator calibration. I run through the calibration OK (or so it seems), but when I try to correct the stigmators (in the Measure tool) and check it in the Focus node, the Thon rings are quite elliptical. Just to clarify, should the same requirements (Gaussian Focus, Eucentric Height and Reset Defocus) apply to the Stigmator calibration as the Defocus calibration? Also, does the beam tilt and stig delta settings influence the calibration greatly? I've set the beam tilt to 0.005 as the 70 micron aperture we're using tends to get in the way at 0.01, and 0.15 for the stig delta. We're working at 115,000 times magnification instead of ~60,000, if that makes a difference. Also, what type of sample is preferred for this calibration? I used one that was mostly carbon film, thinking that it would be good for generating Thon rings, but there wasn't a lot of material there that would cross-correlate well.

BTW, I'm still working off the laptop installation of Leginon, as my plans for installing on another system weren't successful, so I'm not running Leginon 1.1 but rather 1.0.1.

Cheers,
Jamie

I have another question concerning dose and setting up the presets. I'm confused whether the papers on Leginon and other literature recommend a total dose of 10 or 20 electrons per square Angstrom. On the course on Leginon, I think we set up Leginon to get defocus pairs so a low defocus preset had a dose of 10 electrons per square Angstrom and a high defocus preset also had a dose of 10 electrons per square Angstrom. In this set up, the low defocus exposure was obtained first and then the high defocus exposure - resulting in a total dose of 20 electrons per square Angstrom. Are both these images suitable for a single particle reconstruction - or is the second, high defocus exposure just suitable for locating particles in the first, low defocus exposure? In discussions with people in the lab in Leeds, it's been suggested that I should just use a total dose of 10 electrons per square Angstrom?

Also, how important is it to do the part of the dose calibration that requires a Faraday cup? The Leginon manual suggests just using a value of 1 for "screen current to beam current scale factor" if you trust the microscope manufacturer to get the calibration right. Presumably this varies from microscope to microscope even from the same manufacter. (The manual quotes a figure of 0.88 for the Tecnai which is the same or at least a similar type of microscope to ours),

William

Hi, William,

For a defocus pair, the structure information is the same in the two images. Therefore, it makes no sense to use both in the reconstruction. The reliability of the structure is built up by having many different particles of the same structure, not multiple images of the same particle. The second, ef, image is generally for particle picking and other tasks that requires high contrast. In order to gain highest resolution information, en images is sometimes so low contrast that it is impossible to reliably pick particles but the location of these particles can be transferred from what determined from the ef image with a correction on overall shift due to, for example, tiny drift. In such a situation, only en is used for structure reconstruction, and therefore, you only need to keep en image in low dose condition to retain the high resolution information. The dose used on the ef image is arbitrary since it is not used in the actual reconstruction and should not be counted in the 10 e/A^2 limit most people impose. If you find that your first image is unusable because of the low defocus, you should increase its defocus next time you collect data.

As far as the importance of Faraday cup dose measurement, it all depends on how accurate you want your dose value to be. We did a Faraday cup measurement recently and the scale factor changed since our last (several years ago). To us, and I suspect it is true for you, too, consistent values are more important than the accuracy versus the standard. If you get good images that gives you atomic resolution at 15 e/A^2 today, you willl want to use similar dose for your next data collection, right?

Anchi