Non-canonical base pairing

[Guest]

Our RNA structure has G(syn):G(anti) pair.
We wonder how the 3DNA program handles the syn
conformation when it calculates helical parameters.

We have observed negative (!) rise and tilt close to 180 degrees.
Roll and twist seem also to be affected.

I wonder what other parameters are sensitive to anti/syn flip
and if there is a fix that would allow us to calculate sensible
values for the helical parameters.

I would appreciate your comments.

[xiangjun]

Thanks for using 3DNA and posting your question at the forum.

Which version & commands are you using? Could you provide me a working example so I can reproduce your problem? As always, a minimal reproducible example is far more effective than general remarks.

At the base-pair level, 3DNA differentiates M-N vs M+N type pairs:

More generally, when the two bases (M and N) forming a pair have opposing faces, the scalar product of their z-axes is negative. 3DNA designates such pairs, e.g. Watson–Crick A–U, A–T and G–C pairs, as M–N with the ‘–’ symbol used to emphasize the opposing directions. If the M and N bases in a pair share the same face, such as in the Hoogsteen base pair in Figure 2b, the pair is recorded as M+N, with the ‘+’ symbol used to emphasize the similar directions of the bases.

At the dinucleotide step level, it also checks the z-axis directions to calculate reasonable parameters (as in a structure with B-Z junction, implemented in v2.0).

A negative rise is not expected for a reasonable helical structure. Before seeing your structure, however, I don't think I could provide you with a concrete explanation.

Xiang-Jun

[Guest]

Which version & commands are you using?

version 1.5, precompiled for RedHat

Could you provide me a working example so I can reproduce your problem?

I have uploaded to your server the coordinates and the 3DNA output file.

The output was produced by first running find_pair on the coordinates
and then analyze.

I would be grateful for your comments.

Best regards,

W.Rypniewski

[xiangjun]

The two files you uploaded to the 3DNA server did help clarify the issue.

The negative Rise and strange tilt etc parameters are all related to the 6th G+G pair: the syn-G on strand I (X:...8_:[..G]G) has reversed "face" (base orientation) relative to others. The bp frame z-axis direction follows that of strand I base (the leading strand). Have a look at file "ref_frames.dat" following running the command (I am using "Ganti_Gsyn.pdb" as the PDB file name):

Code: [Select]

find_pair Ganti_Gsyn.pdb stdout | analyzeYou will see:

Code: [Select]

...     5 U-A ...
   27.5971    26.4873    48.4652  # origin
    0.7980    -0.6022     0.0226  # x-axis
   -0.5501    -0.7432    -0.3809  # y-axis
    0.2462     0.2915    -0.9243  # z-axis
...     6 G+G ...
   28.7994    26.9129    45.4857  # origin
    0.9194    -0.3494     0.1807  # x-axis
    0.2865     0.9096     0.3009  # y-axis
   -0.2695    -0.2249     0.9364  # z-axis
...     7 A-U ...
   31.2226    28.5432    44.1020  # origin
    0.0211    -0.9974     0.0687  # x-axis
   -0.9847    -0.0326    -0.1714  # y-axis
    0.1732    -0.0641    -0.9828  # z-axis
When two z-axes have opposite directions as in the cases here for the 5th UG/GA (a "blocview" generated structure is attached) and 6th GA/UG steps, the mean z-axis of the middle-frame is not well-defined, leading to the "strange" step parameters you noticed. Nevertheless, this set of parameters can still be used to rigorously rebuild the structure.

In such case, we could reverse the z-axis of the 6th G+G pair, making it pointing in the same direction as other bases along the strand. This will ensure that rise will be positive. However, there are some side-effects with this approach: for one thing, structure "rebuilding" won't work as before; secondly, when taking the 5th and 6th steps separately, there is an ambiguity as to which z-axis should be reversed. For these reasons, I prefer to keep the program as is.

Now if you think about it, the "strange" numbers are actually a good thing: it pinpoints the special structural features in this region, which is not directly comparable to other steps. As with any software tools, it is not just about getting the numbers, but most importantly understanding what the numbers really mean.

The "strange" numbers are directly tied to the choice of base-centered reference frame. The base-pair normal and RC8-YC6 based reference frame as used in SCHNAaP/CEHS (which follows NewHelix) is more intuitive. As a matter of fact, 3DNA also include a utility program "cehs" which calculates authentic SCHNAaP/CEHS parameters. Run it as follows:

Code: [Select]

find_pair Ganti_Gsyn.pdb stdout | cehsand check file "Ganti_Gsyn.outc".

For your structure, please also check the section titled "Global linear helical axis defined ..." from 3DNA "analyze" output. The global parameters based on C1'-C1' vectors make sense to me: it is these C1'-C1' vectors that show the periodicity, not being disturbed by the G+G pairs.

HTH,

Xiang-Jun

PS. The attached image was generated as follows:

Code: [Select]

find_pair Ganti_Gsyn.pdb stdout | analyze
ex_str -5 stacking.pdb s5.pdb
blocview -o -i=s5.png s5.pdb