Incorrect topology assignment

[kogucior]

Dear Dr. Xiang-Jun Lu,

I was browsing prepared by you the G-quadruplex database (DSSR-G4DB) and I
encountered an error.
I think that for "http://g4.x3dna.org/entries/2kqg/index.html" the
"+P+P+P" topology is wrongly assigned, it should be "-P-P-P".
I discovered that maybe this is due to a mis-ordering of G-strands
(#1-#4-#3-#2), where the first plane has a composition:
1 glyco-bond = s --- groove = w - n Major -> WC nts = 4 GGGG A.DG2,
A.DG18, A.DG14, A.DG6

and it should have:

1 glyco-bond = s --- groove = w - n Major -> WC nts = 4 GGGG A.DG2, A.DG6,
A.DG14, A.DG18

It is possible that there are more mistakes. I am curious what they come
from and I am looking forward to the revised version of the database.

[xiangjun]

Hi kogucior,

Thanks for using DSSR-G4DB, and for reporting potential issues of the resource on the 3DNA Forum.

I understand what you mean, and DSSR is performing as designed for this case, even though the underlying convention may be changed via an option. You're right that the "inconsistency" is due to the first G-tetrad. Specifically, it is because of the first G, A.DG2, which is in syn instead of anti conformation. There are solid reasons why DSSR is behaving the way it does, one of which being consistent and systematic, instead of ad hoc.

The ordering of four G's in the first G-tetrad in such cases in a deliberate decision. The clockwise (+) and anti-clockwise (-) directionality of the loops, however, can be revised to reflect the progression of the backbone. Your reported case is a good example and will be taken into consideration in the next major release of the DSSR-G4DB resource.

It is possible that there are more mistakes.

As long the first G is in syn conformation, you will observe the same behavior for other G4 structures in DSSR-G4DB. The results should be self-consistent.

I do not have any papers published on G4 yet. The case here is one point of discussion in a manuscript I am working on.

Best regards,

Xiang-Jun

[xiangjun]

It is possible that there are more mistakes. I am curious what they come
from and I am looking forward to the revised version of the database.

As another concrete example, what do you think the descriptor for PDB entry 6r9k should be?

Xiang-Jun

[kogucior]

As another concrete example, what do you think the descriptor for PDB entry 6r9k should be?

I believe it should be as it is currently described as: +L+P+P.

[kogucior]

As long the first G is in syn conformation, you will observe the same behavior for other G4 structures in DSSR-G4DB. The results should be self-consistent.

I don't understand why the anti/syn conformation is having an effect on this. Theoretically, according to the Webba da Silva formalism [Geometric Formalism for DNA QuadruplexFolding] -P-P-P and +P+P+P topology may have a composition of fully syn or anti G-tetrad.

If the topology determination is based on the syn/anti  composition of 4 guanines in G-tetrad then I would expect +/-(LPP) topology for 2kqg. While the identical structure (2kqh) has a -P-P-P topology.

I expected that the order of the guanines in the planes is based on the order in sequence and the interaction edge (WC or H) with the other adjacent guanines.

[xiangjun]

Hi,

Thank you very much for your feedback.

Technically, the assignment of clockwise (+) vs anti-clockwise (-) assignment is easy to change for these PDB entries with first G in syn conformation, thus the first G-tetard is ordered in Major-->WC direction. The current assignment in DSSR, by default, has considerations with regard to the calculation of rigid-body parameters (twist, rise, etc.). Following your feedback, it makes sense that the assignment of the descriptor is treated separately to follow the formalism of Mateus Webba da Silva.

How about the following for 2kqg:

List of 1 G4-stem
  Note: a G4-stem is defined as a G4-helix with backbone connectivity.
        Bulges are also allowed along each of the four strands.
  stem#1[#1] layers=3 INTRA-molecular loops=3 descriptor=3(-P-P-P) note=parallel(4+0) UUUU parallel
   1  glyco-bond=s--- sugar=.--- groove=w--n Major-->WC nts=4 GGGG A.DG2,A.DG18,A.DG14,A.DG6
   2  glyco-bond=---- sugar=---. groove=---- Major-->WC nts=4 GGGG A.DG3,A.DG19,A.DG15,A.DG7
   3  glyco-bond=---- sugar=---3 groove=---- Major-->WC nts=4 GGGG A.DG4,A.DG20,A.DG16,A.DG8
    step#1  pm(>>,forward)  area=19.15 rise=3.41 twist=21.0
    step#2  pm(>>,forward)  area=12.05 rise=3.47 twist=28.2
    strand#1  U DNA glyco-bond=s-- sugar=.-- nts=3 GGG A.DG2,A.DG3,A.DG4
    strand#2  U DNA glyco-bond=--- sugar=--- nts=3 GGG A.DG18,A.DG19,A.DG20
    strand#3  U DNA glyco-bond=--- sugar=--- nts=3 GGG A.DG14,A.DG15,A.DG16
    strand#4  U DNA glyco-bond=--- sugar=-.3 nts=3 GGG A.DG6,A.DG7,A.DG8
    loop#1 type=propeller strands=[#1,#4] nts=1 C A.DC5
    loop#2 type=propeller strands=[#4,#3] nts=5 CACGA A.DC9,A.DA10,A.DC11,A.DG12,A.DA13
    loop#3 type=propeller strands=[#3,#2] nts=1 A A.DA17

While I am working on this topic, please also check (some of) the following PDB entries, and let me know your thoughts on the descriptors:

Code: [Select]

148d 186d 1bub 1c34 1c35 1c38 1hao 1hap 1hut 1i34 1qdf 1qdh 1rde 201d
230d 2e4i 2f8u 2gku 2hy9 2jpz 2jsk 2jsl 2jsm 2jsq 2kf7 2kf8 2kka 2km3
2kow 2kpr 2kqg 2kzd 2l88 2lod 2lyg 2m6v 2m6w 2m8z 2m91 2may 2mb3 2mbj
2mft 2mfu 2ms9 2mwz 2n2d 4dih 4dii 4lz1 4lz4 4ni7 4ni9 4u5m 5cmx 5ew1
5ew2 5j05 5j4p 5j4w 5j6u 5lqg 5lqh 5mjx 5mta 5mtg 5mvb 5o4d 5oph 5yey
5zev 6ac7 6ccw 6eo6 6eo7 6erl 6evv 6f4z 6fc9 6ftu 6gh0 6gn7 6gzn 6h1k
6ia4 6jkn 6jwd 6jwe 6kfj 6l8m 6l92 6r9k 6r9l 6rs3 6tc8 6tcg 6ycv 6yep
6z8v 6z8w 6z8x 6zx6 7atz 7cv3 7cv4 7ntu
Best regards,

Xiang-Jun

[kogucior]

loop#1 type=propeller strands=[#1,#4] nts=1 C A.DC5
    loop#2 type=propeller strands=[#4,#3] nts=5 CACGA A.DC9,A.DA10,A.DC11,A.DG12,A.DA13
    loop#3 type=propeller strands=[#3,#2] nts=1 A A.DA17

I still have doubts about assigning strand numbers, e.g. loop #1 in my opinion occurs between strand #1 and #2. Until now I thought that the order of the strands is due to their relative arrangement in the space and that the counting starts at the 5'-end and goes counterclockwise.

[xiangjun]

Let's agree to disagree on this point for the time being. The directionality in each G-tetrad is specified unambiguously using either "Major-->WC" or "WC-->Major". The first G-tetrad directs the assignment of the four strands of the G-quadruplex.

Please check the descriptors of the other PDB entries I listed, specifically 2GKU, 2LOD, and 4U5M if not more, including groove widths. Let me know if the descriptors are "correct" in your option. Overall, the strands assignment and the corresponding +/- loop directions follow certain convention. I've added a switch in DSSR Pro to allow for both, as user desire. I've been in direct communication with Dr. Webba da Silva for several years. I may change the current default settings later, when a paper on G4 (including G4DB) is published.

Best regards,

Xiang-Jun

[kogucior]

Ok, I'll check these structures, but it looks like our discussion goes beyond the canonical G-quadruplex structures (think in Webb da Silva's convention), hence the problems.
I look forward to the update if everything will be clearly explain in the publication. However, I hope that the "new convention" will be consistent with eg interpretations of CD spectra.
Unfortunately I am using the basic version of DSSR,  I have to think about buying the PRO version.

[xiangjun]

Ok, I'll check these structures,

Please do. The more concrete examples, the better.

but it looks like our discussion goes beyond the canonical G-quadruplex structures (think in Webb da Silva's convention), hence the problems.

That is correct. Even 'canonical' G4 structures in the PDB have more complicated topologies than the Webb da Silva formalism allows. This is where DSSR can play a significant role: It can be used consistently on all G4s, and any "inconsistencies" are worth investigating further. Either DSSR needs to be improved (which I'd be delighted to do) or the structures themselves have some oddities (which is far more likely).

I look forward to the update if everything will be clearly explain in the publication. However, I hope that the "new convention" will be consistent with eg interpretations of CD spectra.

Do not expect to see a DSSR publication on G4 structures that have "everything clearly explained". For one thing, I have no idea on how a convention can "be consistent with eg interpretations of CD spectra". I simple do not have that expertise. Furthermore, the first DSSR paper dedicated to G4s may not be submitted/published anytime soon.

Unfortunately I am using the basic version of DSSR,  I have to think about buying the PRO version.

Hopefully, the DSSR Basic version and DSSR-G4DB have already been of some help to your project. If you decide to go for DSSR Pro, I may add an option that is tailored to your need.

There are many publications on structural analysis, annotations, and nomenclature of G4. If you find other free or more cost effective options, please let us know.

Best regards,

Xiang-Jun