Symmetry

Submitted by kylelmorris on Fri, 2015-04-24 06:15

Forums

Frealign

Hi all,

Thanks for your time and apologies if I'm just missing something simple!

I am moving over from using Relion to refine a structure with T symmetry. I'm aware that the convention for this symmetry group in Frealign is different to Relion and indeed reconstructions in Frealign are quite peculiar when imposing T symmetry, so I have something wrong. I can't figure out exactly what to do, presumably move my reference volume, but can you suggest what adjustment I should make to fit the conventions used by Frealign when coming from Relion?

Many thanks for your help!

niko Fri, 2015-04-24 15:27

Frealign has the option to enter symmetry matrices directly. Do you have symmetry matrices for the Relion T symmetry? If you would like to post them I would be happy to check them against those implemented in Frealign.

kylelmorris Sat, 2015-04-25 06:47

Thanks for your response Niko, it would be great if you wouldn't mind checking. Please find appended the symmetry operators as output by Relion. Please let me know if I can provide any more information.

$ relion_refine --sym T --print_symmetry_ops
++++ Using symmetry group T, with the following 12 transformation matrices:
R(1)=
1 0 0
0 1 0
0 0 1
R(2)=
-0.5 0.86602539 0
-0.86602539 -0.5 0
0 0 1
R(3)=
-0.5 -0.86602539 0
0.86602539 -0.5 0
0 0 1
R(4)=
-1 0 0
0 0.33333 0.94281
0 0.94281 -0.33333
R(5)=
0.5 0.28867 0.8165
0.86603 -0.16667 -0.4714
0 0.94281 -0.33333
R(6)=
0.5 -0.28867 -0.8165
-0.86603 -0.16667 -0.4714
0 0.94281 -0.33333
R(7)=
0.5 0.86603 0
0.28867 -0.16667 0.94281
0.8165 -0.4714 -0.33333
R(8)=
0.5 -0.86603 0
-0.28867 -0.16667 0.94281
-0.8165 -0.4714 -0.33333
R(9)=
-0.5 -0.28867 -0.8165
0.28867 0.83333 -0.4714
0.8165 -0.4714 -0.33333
R(10)=
0 -0.57735 0.8165
-0.57735 -0.66667 -0.4714
0.8165 -0.4714 -0.33333
R(11)=
-0.5 0.28868 0.8165
-0.28868 0.83333 -0.4714
-0.8165 -0.4714 -0.33333
R(12)=
0 0.57735 -0.8165
0.57735 -0.66667 -0.4714
-0.8165 -0.4714 -0.33333

niko Sat, 2015-04-25 14:41

To use Relion's definition for the T symmetry, you will have to modify one of the scripts in the bin directory in your Frealign installation folder. Using a text editor, open the file

mult_reconstruct_n.com

and replace the line

${sym}                                                  !ASYM symmetry card (I=icosahedral)

with the following 10 line

3
-0.5 0.86602539 0
-0.86602539 -0.5 0
0 0 1
-1 0 0
0 0.33333 0.94281
0 0.94281 -0.33333
0 -0.57735 0.8165
-0.57735 -0.66667 -0.4714
0.8165 -0.4714 -0.33333

This should give you a reconstruction with the Relion's symmetry convention. The symmetry is not used for refinement, only for reconstruction. Therefore, this should be the only change necessary.

kylelmorris Mon, 2015-04-27 06:39

Great, thanks Niko! This does of course work.

If I understand correctly, by editing the file will now always apply this symmetry matrix no matter what I input in the mparameters file. Is there a way that I could change just the symmetry matrix for group T? Maybe this is stored elsewhere and I could edit that file?

Otherwise I could just figure out how to move my reference to fit the Frealign convention, may I ask what the T matrix looks like for Frealign or where I could find it to study it and work it out?

niko Mon, 2015-04-27 09:10

Yes, these changes will now always apply and ignore what you set in your mparameters file. You could simply copy the Frealign bin folder with your modified files in a separate location (for example your project directory) and point to it in the mparameters file with frealign_bin_dir. This way you could leave the original installation unchanged.

I am reluctant to change the definition in Frealign. It aligns one of the 2-fold axis along the z-axis whereas Relion aligns the 3-fold axis along z. As far as I know the Frealign convention agrees with that in the International Tables (see Chapter 10, Table 10.1.2.2, in International Tables for Crystallography, Volume A).