SerialEM Note: More About X,Y Positioning

Author

Chen Xu

Contact

<chen.xu@umassmed.edu>

Date-Created

Dec 12, 2019

Last-Updated

Nov 22, 2020

Abstract

Robust positioning to the target position is critical for high level operation of CryoEM data collection. In this note, I like to share my own version of the latest script to perform X,Y positioning task. And I try to explain every line of the code and ideas behind them as well.

I have spent a lot of time thinking and testing about this. If you have better and different ideas, I’d love to hear from you.

The script lines

The script is fairly short, as shown below. It can be inserted in beginning of a single particle data collection script.

X,Y positioning

 buffer = T
 RealignToNavItem 0
 ResetImageShift 2
 #Copy A $buffer     #comment out if a fixed template in $buffer is used.
 AcquireToMatchBuffer $buffer
 AlignTo $buffer 0 1

Explanations Line by Line

buffer = T

This is to define which buffer will be used to store reference image, a whole image or a cropped area of an image. Buffer after N are all beyond rolling range, thus won’t be pushed out by taking too many images.

RealignToNavItem 0

RealignToNavItem is one of the most important functions in SerialEM, in my opinion. It will bring the specimen stage to a valid map item. It typically uses combination of stage shift and image shift to get the job done. 0 here means to stay in the condition from which the map was created. For example, the map was generated using LD View, and the scope currently is at LD R, the scope will switch to the View mag, beam intensity etc.. After realign is done, it stays in View mag. Argument 1 will bring scope back to R, after routine finishes.

This command line will bring the specimen to the picked item position, with some image shift in the last image of the routine takes, in buffer A.

I should point out that this perhaps reflects one of the most fundamental differences between SerialEM and other data collection software - it doesn’t rely on fixed template at all. As long as an item in a valid map is defined (picked), SerialEM will drive the stage there! For example, for a lacey grid with no regular size holes, SerialEM can go to the target positions precisely wherever they are defined as point items.

ResetImageShift 2

ResetImageShift is to clear out any image shift existing in the system and use stage shift to compensate. Then, there is no image shift, which means beam is straight down on the axis. However, the intrinsic inaccuracy of stage movement makes target being slightly off, more or less.

The argument 2 here means stage will clear the backlash by moving to opposite direction for 0.025 microns as default. This can be very useful to slow down the stage drifting after moving to a new location. Low drift is a very good thing since there is no way to correct drifts accumulated within a frame. This is particularly true if one has to use long frame time on some camera system.

#Copy A $buffer

If not commented out, this line will copy the last image (after realign) in buffer A to a target buffer (T in this case). If one uses a fix image, for example, a cropped hole as reference image, then it should be manually copy into T and leave this line commented out.

AcquireToMatchBuffer $buffer

This is a new command, available in 3.8 beta Dec 10, 2019 built and later. It does two things: 1) take a shot using the exact condition of what in the reference buffer for mag, beam condition, binning, exposure time etc.; 2) make the final image the same size as what in the reference buffer, by cropping if necessary. I used to have to do this in a lengthy script using two functions.

AlignTo $buffer 0 1

Simply align the image in buffer A to reference buffer. This would make the target right on again with image shift. The very last argument 1 means no trimming to any of the source image and reference image. This is needed for UltrAuFoil® Holey Gold Films grids which have very “dark” region of the film.

Other thoughts

1. It is helpful to use large defocus offset for map and realigning, as the contrast is significantly better. On our Krios, we use -300um for View offset (in LD).

2. If offset is more than 200um, it most likely needs High-def Mag calibration. With this, system dynamically interpolates the stage shift matrix which is calibrated using near-focus condition. This makes stage movement much more accurate and robust.

3. If possible, use whole image as “dynamic” template instead of sub-area. Using sub-area such as a single hole is a quick workaround for a grid which has periodic feature and 5-point way of picking points might be not very accurate due to local geometry variation.