• EMAN2
• NegativeStain

Negative Stain Single Particle Analysis

There are some special problems which can arise when processing negative stain data. EMAN2 was primarily designed to work with Cryo data, so you may need to do things a little differently to get good results with negative stain.

• Negative stain is generally limited to ~20 A resolution due to stain penetration and general staining quality. It is also basically limited to showing you an envelope surrounding the protein, though there may be some residual low contrast density present inside the stain envelope with inverted contrast.
• Confusingly for novices, single particle reconstructions of negative stain images can sometimes yield FSC curves which extend to subnanometer resolution, even when they are clearly not exhibiting any secondary structure features. This is due to a misunderstanding of what the FSC curve is actually telling us. In this case, it simply means that the sharpness of the stain envelope is better than 10 A. ie - even though it isn't representing high resolution features, it is doing so very self-consistently, resulting in a sharp edge, which corresponds to "high resolution". Such FSC curves are meaningless (or at least not useful in assessing the features you can expect to see).
• There is very little point in collecting large numbers of particles in stain, due to the inherent resolution limit. If your reason to look at large numbers of particles (more than 2-3000, less with symmetry) is to look at variability, just be warned that the conditions in negative stain are quite far from the cryo conditions which are believed to mimic solution conditions. That is, I would be concerned about trusting any 'variability' results in stain. If you are at that point, it is time to move to ice.

• When using the "Import Micrographs & Est. Defocus" item in the project manager (e2rawdata from the command line), do NOT specify the X-ray pixel filter, which eliminates histogram outliers, assuming they are detector artifacts. In negative stain, the noise levels are so low that this detection often fails and zeroes out high contrast pixels which should be retained.

Normally people who know negative stain well, and have nicely stained images will collect their images at or near Scherzer defocus (ie - very close to focus). This minimizes visible CTF artifacts in the image, and since most of the contrast provided by negative stain is amplitude contrast, there is no need to defocus. Other people choose (for various reasons) to image far from focus as if they were working in Cryo.

If working close to focus (<0.5 microns)

• CTF autofitting will not work properly with default settings, but it is still necessary to do CTF correction for the single particle processing pipeline to work.
• If you have fewer than 5 micrographs, you must process CTF manually, as autofitting requires a minimum of 5 to proceed.
• Amplitude contrast should be set to 60-90% (I generally use 80, specific value probably doesn't matter)
• The default CTF autofitting settings do not extend very close to focus, and will not work properly unless ~2 CTF minima are clearly visible in the power spectrum. You have two choices:
  • Adjust the autofitting min/max defocus to span a very narrow range of defocuses near what you used, eg 0.1-0.2 microns. Since you won't be achieving high resolution anyway, a precise value is not important. Autofitting may still fail, so you should check the results with the interactive CTF fitting interface.
  • Alternatively, you can just manually fit the defocus to some reasonable value. It is a bad idea to give all of the images exactly the same defocus, but if you put them in a range (say 0.1 - 0.15 microns), it should be fine. Don't forget the "compute structure factor" and "generate output" steps.
• Negative stain images will normally show as white particles on a darker background. This is the correct contrast for EMAN, so if you see this, there is no need to 'invert' your data at any stage.
• When making class-averages, both in 2-D and as part of 3-D reconstruction, you need to change the "classaverager" parameter to "mean". Otherwise your classes may get overfiltered due to confusion about CTF.

If working far from focus (anything >0.5 microns)

• CTF autofitting may work properly, but should be checked manually with the GUI, particularly if defocus <1 micron. Performing CTF processing is required, even if you cannot achieve high resolution.

• If you have to adjust CTF parameters manually, rerun autofitting after saving updated defocus values. It will use your values as hints and should not wander back to incorrect defocus values.
• If you have fewer than 5 micrographs, you must process CTF manually, as autofitting requires a minimum of 5 to proceed.
• Amplitude contrast should be set to 60-90% (I generally use 80, specific value probably doesn't matter)
• Negative stain images will normally show as white particles on a darker background. This is the correct contrast for EMAN, so if you see this, there is no need to 'invert' your data at any stage.
• When making class-averages, both in 2-D and as part of 3-D reconstruction, you need to change the "classaverager" parameter to "mean". Otherwise your classes may get overfiltered due to confusion about CTF.