Chapter 2: Autoindexing and Cell Refinement


Step 8: Autoindexing


Step 8: Autoindexing

To start autoindexing go back to the HKL-2000 window, click the Index tab. Set the Resolution Limits (the default is to the edge; changing it later is also ok), then autoindex by clicking the index button. Keep the P1 space group for now and close the Bravais Lattice window. Check to make sure the green autoindex circles on the frame image line up with real spots by examining the Image Display window and toggling on and off the update pred button. Click on set blind region and define the beamstop shadow.

How do I define the beamstop shadow and why should I do this?

It is important to define the area of beamstop shadow. This is because unless the beamstop shadow is defined explicitly, Denzo may fail, under certain circumstances of high background, to reject potential reflections, which are in fact obscured by the beamstop. These few inaccurately measured “reflections” will severely degrade the quality of your anomalous signal. To define the beamstop shadow, keep the Image Display window open and open up the Index. Click on Set Blind Region and then click the top large green box in the new window that appears (Figure 23). Now go to the Image Display window and use the left mouse button to define the four corners of the rectangle that corresponds to the beamstop arm (obviously, this does not apply to mylar film beamstops that leave no arm shadow). The blocked out region will appear red. If you don’t like the results you can repeat the process and after selecting your four new vertices the original rectangle will be replaced by your new choice. Similarly, to define the beamstop circle click on the big green circle, then move to your Image Display window and double click on the center of the beamstop. After this you can move the mouse to enlarge or reduce the blind region. You don’t have to hold down a mouse button to do this. When you are satisfied, click once with the left mouse button and a filled red circle will appear. Going back to the Blind Region dialog box, click apply both. If you don’t like your selection, click Clear Current. If you want to begin integrating, click apply current. If you want to store these shadows permanently, click update site and later this information will be forwarded to your Site Configuration, so you will be able to use that for all subsequent processing’s. Click close, when you are done.



Figure 23. The Blind Region definition window and image with assigned blind region

What does the blue highlight in the Sets Pending box mean?

If you have two or more data sets that you are integrating, then these sets will appear in the Pending Sets box in the upper left corner of the Index page. You can click on a set and it will be highlighted with blue background (Figure 24). This is the set that is currently refined. Additionally, when one set is highlighted, you can change the parameters of the integration for that set, while leaving the parameters for the other set(s) unchanged. For example, set 1 may be set for processing to 3 Å resolution, while you may want to process set 2 to 2.5 Å. Similarly, perhaps the Spot Size is 0.3 for the first set, but 0.25 for the second set. Or perhaps the 3D Window is 5 frames for the first set but only 3 for the second. The Pending Sets box allows you to toggle between the pending data sets in preparation for integrating them in the same run. You can also accomplish the same thing by going to the Summary tab page and editing the parameters for each set (Figure 25).


Figure 24. The Pending Sets box


Figure 25. The summary information about processed sets

Why would I want to put in any other resolution limits?

Before autoindexing, the only parameter you will want to consider changing from the default values is the Resolution. If your data is good to the edge of the plate, then leave the default settings, as they are (Figure 26). In other cases, however, for example if you have ice rings in your data, or the data are weak beyond a certain resolution, you are better off changing the Max resolution value to something more reasonable. For example, if you have an ice ring, try setting the Max value to 4 Å. If you have a secondary lattice which is more pronounced at higher resolution, try indexing at low resolution.


Figure 26. The Resolution limits box

At what point should I be concerned with the Refinement Options?

Not yet. The initial parameters are those of fit basic and these are appropriate for autoindexing and first refinement step. In general, thanks to eigenvalue filtering, you can integrate your data with fit all and any parameters that are too highly correlated will be restrained by the program (Figure 27). The days of selective manual fitting and fixing are over. The program almost always does a better job.

*    Manual intervention may still be appropriate in cases of multiple lattices or other crystal pathologies.

Indexing keeps flashing. How long is this supposed to take?

For autoindexing, the computer has to read in all the frames of the 3D Window for which peak searches have been performed and then do the calculation. Generally autoindexing takes less than a minute, but is roughly proportional to the number of reflections to be considered for autoindexing.


Figure 27. The Refinement Options box

What if there aren’t any green circles? What if they don’t line up on real spots?

The green circles are a subset of the initial peaks selected in the peak search (Figure 28). This is the subset that has been used in the autoindexing. How are these green circles selected? First, the image frame is read into memory by the autoindexing program and the peak search coordinates are mapped onto this image. The autoindexing program then decides whether in fact these coordinates represent usable peaks or not. The criteria for usability include falling within the resolution limits specified, high enough signal to noise, sufficiently low background (which helps to eliminate spots in ice rings and overlaps), and identification of a lune. If there aren’t any green circles after autoindexing, one or more of the following problems may be the cause:

  1. The Peak Search was done on a different frame from that being used to autoindex. In this case you should have already gotten the error message “Error – Peaks none.” The remedy is to make sure these are the same. The rightmost column of the file peaks.file, generated during peak searching, tells which frame the peak search peaks were derived from.

  2. None of the spots are strong enough to qualify for use in the autoindexing routine. The remedy is to take a longer exposure, pick more peaks on the frame or on several frames, or lower the Weak Level, which is normally set at 5 σ. Lowering the weak level is a dangerous option, however, and is not recommended because weak reflections are not high quality reflections, and therefore may be noise.

  3. Rejection criteria for background. Denzo assesses the quality of the background of each spot against criteria for slope and uniformity, and if certain fraction of the pixels in the background fails the test, then the whole spot is rejected. This is controlled by the keyword Reject Fraction. This default value for this rejection is 0.75 i.e. the 75% of the background pixels must be acceptable. If you have an ice ring, you can increase Reject Fraction until most of the spots in the ice ring are not longer picked up. Useful increments are 1-2% at a time. Do this while you are looking at the display, just as you did for spot shape and mosaicity.

  4. There weren’t enough spots to describe a lune (sparse lattices, low mosaicity, and low resolution). The remedy is to pick more peaks or include more frames in the peak search.

 A further discussion of autoindexing failures can be found further down.



Figure 28. The Zoom window with image after autoindexing




Displaying your image


Table of Contents


Lattice parameter refinement