1MZP

[cigdem]

Hi Xiang-Jun,

I was looking at the structure 1MZP and I noticed that DSSR does not notify its k-turn motif. However, if I use --nested option, it gives the k-turn information.
I noticed that it is considered as having 1 internal loop (with 1 1 unpaired bases) and a 6-way junction. When I look at the 2D structure, I see two internal loops and a pseudoknot. This is a little confusing for me.

[xiangjun]

Hi Cigdem,

You've brought up an interesting point on the identification of junction loops in DSSR. By default, pseudo-knotted stems/isolated canonical pairs (the A13–U43 Watson-Crick pair in 1mzp) are also taken into consideration in deriving the 6-way junction loop. See the attached 2d-diagram in linear-arc presentation, and make use of the corresponding DSSR output (excerpted below) to figure out the 'route' of the 6-way junction:

Code: [Select]

List of 1 junction
   1* 6-way junction: nts=33; [6,1,2,10,2,0]; linked by [#-1,#-2,#2,#3,#-2,#2]
     summary: [6] 6 1 2 10 2 0 [B.6 B.50 B.13 B.43 B.45 B.20 B.23 B.32 B.43 B.13 B.16 B.49] 1 1 5 3 1 5
     nts=33 GCGUAGGAUACGGAGCGCCGGUGAAAUAUAGCC B.G6,B.C7,B.G8,B.U9,B.A10,B.G11,B.G12,B.A13,B.U43,B.A44,B.C45,B.G20,B.G21,B.A22,B.G23,B.C32,B.G33,B.C34,B.C35,B.G36,B.G37,B.U38,B.G39,B.A40,B.A41,B.A42,B.U43,B.A13,B.U14,B.A15,B.G16,B.C49,B.C50
       nts=6 CGUAGG B.C7,B.G8,B.U9,B.A10,B.G11,B.G12
       nts=1 A B.A44
       nts=2 GA B.G21,B.A22
       nts=10 GCCGGUGAAA B.G33,B.C34,B.C35,B.G36,B.G37,B.U38,B.G39,B.A40,B.A41,B.A42
       nts=2 UA B.U14,B.A15
       nts=0

The DSSR NAR paper (including the User Manual) defines isolated canonical pairs, pseudo-knots, and junction loops etc., and contains quite a few examples. You may find it helpful to revisit the paper and documentation.

In its current definition, DSSR takes k-turn as a special case of internal loops. Since the presumed k-turn components are also already included in the 6-way junction loop, it is no longer considered to be in an internal loop. Removing pseudoknots with the option --nest as you did reveal the internal loop, and thus the k-turn.

I'm sure you are familiar with the paper "The k-junction motif in RNA structure" by the Lilley group, where they introduced the concept of k-junctions. I believe the k-turn in 1mzp belongs to a k-junction (as broadly defined in DSSR). You may want to try their computer program described in the k-junction paper:

We therefore wrote a computer program that would search within RNA structures for two helices with a relative inclination that was similar to that of the geometry of the C and NC helices of a standard k-turn.

...

We wrote a computer program in Python 2.7 to analyse relative orientation of segments of double helix in coordinate files of RNA structures downloaded from the PDB. A search pattern was defined by calculating the relative coordinates of helix segments from the C and NC helix of known k-turn structures. This structural criterion was then applied to other RNA structures to search for sections in which two helical segments had a relative orientation that was similar to those of a conventional k-turn. For all helix segments we searched for nearby candidate helical segments that matched the search pattern. Using this approach we observed that the A. thaliana TPP riboswitch (PDB code 3D2G) contained two helices with a strong structural similarity to the k-turn geometry.

Please let me how it goes. The software from the Lilley lab may well fit your needs. If needed, I may decide to expand DSSR's k-turn definition to include junction loops. This will surely reveal more k-turns by default than DSSR currently detects.

Best regards,

Xiang-Jun

[cigdem]

Hi Xiang-Jun,

Thank you for the explanation. It makes sense, if DSSR defines a junction loop by also considering the pseudo knotted pairs then it won't define a k-turn motif since there is no internal loop.

Yes, I am familiar with Lilley's k-junction definition. It is found in TPP riboswitches I guess.
On my side, I am also not considering the k-junctions for now. Thanks a lot!

Sincerely,
Cigdem