resolution and refinement

Please use this forum to post questions on general small molecule X-ray crystallography from theory to practical.
Posts: 2
Joined: 09 Nov 2017, 18:46

resolution and refinement

Post by hassiba » 09 Nov 2017, 21:04

hi forum members
I have a structure when the DMSO is in two positions with occupancy.

how to refined rather than fixed in final refinements.

and another think about The cut off of sin(theta)/lambda = 0.6, how to do to use more data.

thanks in advance
Last edited by johnewarren on 10 Nov 2017, 08:24, edited 1 time in total.
Reason: Posted in wrong section of forum and into unrelated topic

User avatar
Rotating Anode With Optics
Posts: 207
Joined: 26 Jan 2012, 16:04
Location: Dublin

Re: resolution and refinement

Post by Helge » 10 Nov 2017, 09:16

As for the DMSO: I am a little confused. What do you want to refine? The occupancies? That works as follows: Set the occupancy of one part to 21.0 and of the other one to -21.0. Then add "0.5" at the end of the FVAR line.

As for the resolution: DO you have any line in the instruction file that curtails your resolution, i.e. either OMIT or SHEL? If not, then all data that has been measured is included in the refinement (as it should be), so the only way to get higher resolution data is to measure it anew.

Posts: 2
Joined: 09 Nov 2017, 18:46

Re: resolution and refinement

Post by hassiba » 15 Nov 2017, 14:48

hi forum members
I submit my article in Acta Crystallographica E
and they give me these corrections
(If the DMSO is in two positions with occupancy these should be refined rather than fixed in final refinements.

Most importantly, you should use more data. The cut off of sin(theta)/lambda = 0.6 should be regarded as the absolute minimum value. You have measured beyond this point so should use these data. I suggest a command such as SHEL 50 0.78 is appropriate here. Routinely trimming data to SHEL 50 0.84 is inadvisable; while this gives lower R-factors it also tends to give poorer precision.

Here you can use the recorded data but do not trim it to the resolution of 0.84 but 0.78. Similarly, be careful with the OMIT commands as described below.

In a similar vein, the OMIT command has been overused here. Some programs such as OLEX2 suggest the omission of a huge raft of data and much of the time this is unjustifiable.

OMIT is used to remove a very small number (one or two) of very badly fitted reflections.
OMIT is also used to remove those reflections that are incorrectly measured eg they have grossly underestimated intensity because of the beam stop or the low angle. I suggest these are appropriate to OMIT here:

OMIT 1 0 1
OMIT 0 1 2
OMIT 0 0 2
OMIT -1 0 1
OMIT 0 2 0
OMIT 1 1 0
OMIT 0 1 1
OMIT 0 1 3
OMIT 0 2 1

Reading the SHELXL.LST file will help in choosing reflections. OMIT those that have very small or negative intensities and are at low angle).

Please repeat the refinements with more data included and update the paper.
thanks in advance

User avatar
Olex2 Boffin
Posts: 72
Joined: 12 May 2008, 09:40

Re: resolution and refinement

Post by HorstPuschmann » 17 Nov 2017, 10:22

Just a comment: Olex2 doesn't "suggest the omission of a huge raft of data" anywhere. I have no idea where that referee would have got that idea from.

What Olex2 *does* offer is the facility to examine the most disagreeable reflections in detail -- i.e. you can inspect each occurrence of the reflection and then decides which of these you might want to exclude without having to throw out the entire reflections.

In any case, I agree that the OMIT list reproduced here is strange -- it looks like something hasn't gone quite to plan in the data processing step somewhere.

I also have a problem with the notion that the occupancy factors of disordered parts should be refined as opposed to fixed. There are many cases where the refined occupancy ends up as something like 48/52. I would much rather fix these at 50/50. The fact that the embedded CIF file contains your complete model, as well as the full data, gives anyone a chance to check your assignment if they must.

It sounds like this structure might be a good candidate to turn into a video -- if you can share the data, please do!

Sealed Tube With Optics
Posts: 96
Joined: 08 Mar 2011, 22:50

Re: resolution and refinement

Post by dkratzert » 17 Nov 2017, 16:38

Hi Horst,

I suspect with "suggest the omission of a huge raft of data" the option "OMIT all reflections where |Error/esd| > than [ 5 ]" is meant. The default is 5 which I also find way too low. In some of my structures this would omit over 50 reflections. I think the default here should be the same where you color the error/esd in red.


Return to “General SMX”