defining a local helix axis

[mark]

I am analyzing a protein-DNA complex in which the DNA is somewhat underwound and may have several small kinks. I wanted to calculate the helical trajectory along the DNA helix and wondered if I could use 3DNA to do this. 3DNA output gives coordinates for the origin of each base pair (Ox,Oy,Oz) as well as coordinates for the position (Px,Py,Pz) for each dinucleotide step. Would one or the other of these coordinate sets be more appropriate to use, or are neither of these a good measure of the helix axis?

Thanks!

[xiangjun]

Thanks for using 3DNA, and posting interesting questions on defining local helix axis and quantifying the kinks in a DNA structure.

3DNA outputs many structural parameters including "Origin (Ox, Oy, Oz) and mean normal vector (Nx, Ny, Nz) of each base-pair in the coordinate system of the given structure", and "Position (Px, Py, Pz) and local helical axis vector (Hx, Hy, Hz) for each dinucleotide step", as shown in file bdl084.out distributed with 3DNA. In idealized cases, e.g., where a perfectly regular B-DNA fragment connects to a perfectly regular A-DNA fragment, the local helical axis vector should do the trick. However, in normal cases, e.g., for your protein-bound DNA, local distortions make local helical axis associated with each dinucleotide quite irregular (see an worked example below), and as always, one needs to be careful in drawing conclusions from using it.

Local kinks in protein-bound DNAs are normally quantified using roll angle in the literature. See, for examples, the two Chen et al. papers: http://www.ncbi.nlm.nih.gov/pubmed/11724532 and http://www.ncbi.nlm.nih.gov/pubmed/11724533.

Also, you may want to check Curves, which has been used frequently in the literature for quantifying DNA curvature. It is to the users' advantage to have a choice for comparisons before jumping to conclusions. I have tried to build a bridge between 3DNA and Curves to make Curves users' life a bit more straightforward: find_pair have an -c option to generate input to Curves from a PDB file.

Since you are inquiring about "defining a local helix axis", the following message (slightly edited), which I communicated with an acquainted 3DNA user back on January 5, 2007, could be useful and/or more directly relevant.

> [2] my point is that the global axis is NOT displayed in the Raster3D mode.
> Nothing changes when I run frame_mol and then alc2img with the -g option.

[...]That 3DNA forum has been created for general discussions, so that the
community knows that 3DNA is still under active support and further
development.

3DNA certainly has more functionalities than described in our 2003 NAR
paper, and in reality, it is the details that count.

The simplest way to answer your 2nd question is by following an example
such as below:

Code: [Select]

find_pair bdl084.pdb stdout | analyze
rebuild bp_step.par bdl084.alc
rasmol -alchemy bdl084.alc
frame_mol -a -g -m -6,7 ref_frames.dat bdl084.alc bdl084_new.alc
rasmol -alchemy -noconnect bdl084_new.alc
alc2img -a -g -l bdl084_new.alc t.ps
gv t.ps
alc2img -r -a -g -l bdl084_new.alc t.r3d
render < t.r3d > t.avs
display t.avsThe global axis is displayed in t.avs, or t.ps with x-, y- etc labels.

Note: The above "recipe" works with 3DNA v1.5; One should use RasMol 2.6.x, not v2.7.x (a current version of Jmol is also fine) to display ALCHEMY format files properly; The command 'display' is from the ImageMagick package. See 3DNA 2008 NP paper for more examples.
[hr:611u016p][/hr:611u016p]
HTH,

Xiang-Jun

[mark]

Thanks Xiang-Jun,

I simply put the base-pair origin coordinates (Ox,Oy,Oz) into a pdb file and diplayed those in Pymol along with the DNA coordinates. The origin coordinates do suggest regions of kinking but I will go back and also verify this with Curves and also with reference to the roll angles as you suggest.

Thanks again,
Mark