Definition of Helix Form

[lvelve0901]

Hi, xiangjun,

How is everything going?

In DSSR helices section, there is helix-form part including 'A', 'B' or 'Z' for the common A-, B- and Z-form helices, '.' for an unclassified step, and 'x' for a step without a continuous backbone.

My question is:
How do you identify the the helix form? Which parameter did you use to define the 'A' 'B' 'Z' form?

Best,
Honglue

[xiangjun]

Hi Honglue,

Still working on what I promised (i.e., groove widths etc in JSON) -- it is not as straightforward as I originally thought, but I am making progress...

In the DSSR output, the classification of a dinucleotide step into 'A', 'B' or 'Z' is based on new scheme. The basic idea is on comparisons to 'reference' structures, even though technically more involved. The details will be reported later.

RNA helices are predominately in A-form, and DSSR targets on RNA users to begin with. However, DSSR also works for DNA where A-, B- and Z-forms all exist. The classification summary would be useful especially when A/B-form transition happens, as in 1tc3.

Best regards,

Xiang-Jun

[lvelve0901]

Hi Xiangjun,

Could you please tell me when you will report the detail of helix form definition in dssr? My advisor asked me some questions in terms of how dssr identify helix form so I would like to know more details of this.

This is not urgent so please take your time.

Thank you so much.

Best,
Honglue

[xiangjun]

Hi Honglue,

Thanks for your continued interest in DSSR's classifications of A-, B- and Z-helical forms. There are actually many such details in DSSR which I take as 'experimental' and are not published in the 2015 NAR paper or documented in the user manual.  Some of them will certainly end up in new publications, but no timelines.

As for DSSR reported helical forms, do they make sense? Or did you notice anything peculiar? I'd welcome your feedback.

Xiang-Jun

PS. I attended your advisor seminar (very informative) early this month at Rutgers. I saw your name in his acknowledgment list...

[lvelve0901]

I actually did some work to benchmark whether 3DNA did a good job to identify the helix form.

In our lab, we have an in-hosue database of all the DNA stem and RNA stem structures from the entire crystal structures labeled "Protein#DNA" and "Protein#RNA" deposited in RCSB with resolution under 4Å. Yes, I use DSSR to generate stem structures for each PDB.

Then I build fiber idealized B form DNA and idealized A form RNA using 3DNA. If you type

fiber -m

It will generate a list of different nucleic acid model, I pick the number 4 for B-DNA and number 20 for A-RNA.

[hs189@summer:Plot] fiber -m
Fiber data in directory: /home/hs189/X3DNA/fiber/

id#  Twist   Rise        Structure description
   (degree) (Angstrom)
-------------------------------------------------------------------------------
 1   32.7   2.548  A-DNA  (calf thymus; generic sequence: A, C, G and T)
 2   65.5   5.095  A-DNA  poly d(ABr5U) : poly d(ABr5U)
 3    0.0  28.030  A-DNA  (calf thymus) poly d(A1T2C3G4G5A6A7T8G9G10T11) :
                                        poly d(A1C2C3A4T5T6C7C8G9A10T11)
 4   36.0   3.375  B-DNA  (calf thymus; generic sequence: A, C, G and T)
 5   72.0   6.720  B-DNA  poly d(CG) : poly d(CG)
 6  180.0  16.864  B-DNA  (calf thymus) poly d(C1C2C3C4C5) : poly d(G6G7G8G9G10)
 7   38.6   3.310  C-DNA  (calf thymus; generic sequence: A, C, G and T)
 8   40.0   3.312  C-DNA  poly d(GGT) : poly d(ACC)
 9  120.0   9.937  C-DNA  poly d(G1G2T3) : poly d(A4C5C6)
10   80.0   6.467  C-DNA  poly d(AG) : poly d(CT)
11   80.0   6.467  C-DNA  poly d(A1G2) : poly d(C3T4)
12   45.0   3.013  D-DNA  poly d(AAT) : poly d(ATT)
13   90.0   6.125  D-DNA  poly d(CI) : poly d(CI)
14  -90.0  18.500  D-DNA  poly d(A1T2A3T4A5T6) : poly d(A1T2A3T4A5T6)
15  -60.0   7.250  Z-DNA  poly d(GC) : poly d(GC)
16  -51.4   7.571  Z-DNA  poly d(As4T) : poly d(As4T)
17    0.0  10.200  L-DNA  (calf thymus) poly d(GC) : poly d(GC)
18   36.0   3.230  B'-DNA alpha poly d(A) : poly d(T) (H-DNA)
19   36.0   3.233  B'-DNA beta2 poly d(A) : poly d(T) (H-DNA  beta)
20   32.7   2.812  A-RNA  poly (A) : poly (U)

I know that the 3DNA identify the helix form in a dinucleotide step so I generated two base pair long idealized B-DNA and A-RNA to align the coordinate of the stem structures I generated using only backbone and sugar heavy atom and yielded an alignment RMSD for each dinucleotide step in my database.

Here is the result:

My RMSD cutoff is 2Å.

Protein#DNA
Total number of entries (dinucleotide step): 97366
Number of entries with RMSD (> 2Å) but 3DNA think it is B form: 49
Number of entries with RMSD (< 2Å) but 3DNA think it is X form (ambiguous): 29702
The rest of entries is 3DNA agree with my RMSD cut off.

Protein#RNA
Total number of entries (dinucleotide step): 56530
Number of entries with RMSD (> 2Å) but 3DNA think it is A form: 0
Number of entries with RMSD (< 2Å) but 3DNA think it is X form (ambiguous): 22893
The rest of entries is 3DNA agree with my RMSD cut off.

I think 3DNA basically did a good job considering the number of entries that excess the RMSD cutoff but 3DNA think it is A/B form among the entire PDB.
I am just wondering does 3DNA also simply use coordinate alignment to identify the helix form?

Best,
Honglue

PS. these data is under publication so I am not sure if I can provide further details but I will try my best to give you as much detain as you want.

[xiangjun]

PS. these data is under publication so I am not sure if I can provide further details but I will try my best to give you as much detain as you want.

I understand. Note also that the A-, B-, and Z-form classification algorithm implemented in DSSR is unpublished yet, as I mentioned previously. DSSR contains many undocumented features, which, if elaborated in combination with a survey of the PDB, could lead to solid publications. I've limited resources, and have to focus on the essence of DSSR as a whole. I'm glad to see that some users indeed have taken advantage of specific features of what DSSR has to offer. With collaborations, I may help to dig into some specific topics more thoroughly and efficiently.

I've no comment on your unpublished results. Since you're also using 3DNA v2.x, it helps to check the A- and B-form classification results from the analyze program. The fiber also contains the -rna option.

Xiang-Jun