I have set up a batch job for a set of images taken from one crystal at CHESS. I indexed the first one with autoindex and wrote a shell script. I understand that with the Fuji scanner you can give sector 1 to n instead of giving the line @refine n times. But then some images did not refine because the beam position shifts a lot. I picked the beam position in the display and wrote the following: sector 1... @refine x beam ... y beam... @refine....So I still have one @refine per image but can alter the beam position before the next image. Is this the best way to do it (it works) or is there a better way?

This will work.


Autoindexing gives almost correct cell constants, but the predictions are somewhat off, so the refinement does not kick in. I have 60 frames, but the data is weak and only to 4Å.

There are some improvements to refinement algorithm that make it more stable, however the problem is likely be somewhere else. One likely explanation that would agree perfectly with your description is misindexing of lattice by one index. The solution in such a case is to input correct (or close to correct) values of x beam and y beam.


Is it the sector n to m command that sets up the do-loop for data integration? Therefore, if one puts this command at the beginning of the auto.dat section, one will get independent indexing of each image; if it goes at the beginning of refine.dat then one indexing is used and each image is simply made to fit as best it can. Right?

No. It does not matter where you put the sector n to m command as long as it is before the first go statement. The do-loop starts at start refinement and ends at calculate go


When I ran a batch job to process 46 frames, I noticed that the crystal rotx was not updated with the oscillation step during the time of each job. For example, it I start with crystal rotx of 32.167, after 5 frames are processed, the value of the crystal rotx is about the same, 31.469. However, the oscillation start and end are updated each time. How can we understand this?

That you collected data starting each image at a different oscillation start, the way everybody collects his/her data.


Last time you said that crystal rotx is defined relative to spindle angle for which oscillation angle is 0 and it should not change when advancing to the next frame (within a batch job). However, when I processed the data frame by frame manually, the crystal rotx values changed each time at the end of the refinement no matter what value was input before the process. Should the crystal rotx values be different for the different frames (only different by an oscillation step, saying 0.5) or should they be the same? What I have seen is that the crystal rotx values are the same for each frame when the job is submitted as a batch one, while they are quite different (not within an oscillation step) when the data is processed manually frame by frame. If the crystal rotx values are different for every frame, how can one include partial reflections during the process of scaling?

What counts is only sum of the oscillation start + crystal rotx. It seems that you are inconsistent in how you describe crystal orientation using different procedures.


In the sample input of Scalepack ( you sent me a while ago, there are statements for postrefinement. After the postrefinement do you suggest a reintegration?

It depends. If mosaicity from postrefinement is higher than that set in Denzo, then definitely yes. If you want to precisely characterize the detector, you may wish to rerefine and reintegrate the data with unit cell parameters fixed to postrefinement results. Detector parameters refined in Denzo will then be much more precise. Otherwise the gain from reprocessing is very small.


What does start refinement do?

Start a loop that finishes at calculate go. Loop range is defined by sector 1 to 50. This will execute the loop 50 times.


Comments: Zero s often comes from too-small (or absent) profile fitting radius. Increase the value and judge it on the display (click top left button) .. For small molecule crystal spot density is smaller and larger profile fitting radius may be optimal.