DSSR multiplets

[febos]

Dear Dr. Lu,

At the moment I study RNA base triples found in RNA-containing PDB entries using DSSR. I and my students have created a new classification of base triples with respect to secondary structure environment of their nucleotides and now we are planning to implement the classification within our database's web-interface.
We thought it would be helpful to include the 3D visualization of an arbitrary set of aligned base triples and while working on that problem I've noticed that DSSR stores aligned base triples within its output file named dssr-multiplets.pdb.
Could you please tell me how you align multiplets in DSSR? Unfortunately I couldn't have found this information in DSSR tutorials.

Thanks in advance.

Respectfully,
Eugene

[xiangjun]

Hi Eugene,

Welcome back! I'm glad to hear of your continued interest in using DSSR in your project(s).

Each multiplet identified by DSSR is outputted in sequential order. The multiplets are ordered by the number of nucleotides, and then by sequential numbers. I do not know if that's what you mean by "aligned". You may find the section "Summary of structural features of ** nucleotides" relevant.

As a side note, (by default) each multiplet has been reoriented in the 'top' view to illustrate the overall planarity of the nucleotides.

HTH,

Xiang-Jun

PS. DSSR contains many undocumented features. The DSSR User Manual is over 90 pages, and that may be already too long for general users.

[febos]

Dr. Lu,

I'm sorry for not being clear. By "aligned" I mean that if I take several base triples from an arbitrary dssr-multiplets.pdb file and look at them in Jmol - they appear aligned. So my question is about processing coordinates of the nucleotides - the coordinates of nucleotides in dssr-multiplets.pdb are different from initial PDB entry. Could you please explain what transformation DSSR performs in there?

Respectfully,
Eugene

[xiangjun]

OK. Now I see what you meant by "aligned".

The algorithm in DSSR for setting the 'planar' view of a multiplet works as follows:

• The base reference frame of each nucleotide in the multiplet is known. See for example
  

  x3dna-dssr -i=1ehz.pdb --json | jq .nts[0].frame

• The z-axis of the multiplet is the mean of the z-axes of all base frames, using the first (z1) as reference. If the z-axis (z2) of the other base is anti-parallel to z1 [i.e., dot(z1, z2) < 0], z2 is reversed before taking the average. The z-axis of the multiplet is then normalized.
• The x-axis of the multiplet is defined similarly, and corrected for orthogonality with the z-axis.
• The y-axis is defined as cross(z_axis, x_axis) for a right-handed reference system.
• The origin of the multiplet is the geometric average of the origins of the base frames.

The multiplet is then transformed to the coordinate frame defined above. By definition, it has the 'top' view to show the planarity of the multiplet. Note that the x-axis and y-axis of the multiplet could be defined differently. They correspond to different rotations around the z-axis in the multiplet plane.

Over the development of DSSR (and 3DNA), I've tried different approaches in setting the x-axis and y-axis of the multiplet. No users have noticed/reported such subtle changes. If you'd like to have full control over the multiplet coordinates, please try the DSSR --frame option (see the DSSR User Manual).

For completeness, DSSR also provides the --raw-xyz that outputs the original coordinates as in the input PDB/mmCIF file, without any further transformations.

HTH,

Xiang-Jun

[febos]

Dr. Lu,

That's it! Thank you!

Respectfully,
Eugene