Global linear helical axis calculation in 3DNA

[kateryna]

Dear 3DNA users,

I'm trying to understand how the global linear helical axis is calculated in 3DNA. I've started with building DNA decamer (CGTACGTACG)2 in standard B-form using FIBER and analyzing it with 3DNA. Here is the section of output concerning global linear helical axis:

Code: [Select]

****************************************************************************
Global linear helical axis defined by equivalent C1' and RN9/YN1 atom pairs
Deviation from regular linear helix: 3.38(0.00)
Helix:    -0.000  -0.000  -1.000
HETATM 9998  XS    X X 999      -0.000  -0.000   0.000
HETATM 9999  XE    X X 999      -0.000  -0.000 -30.375
Average and standard deviation of helix radius:
      P: 9.23(0.00), O4': 6.15(0.00),  C1': 5.68(0.00)The phrase "Global linear helical axis defined by equivalent C1' and RN9/YN1 atom pairs" is not quite clear for me. As I understand it, in each base pair four atoms  (RC1', YN1,YC1',RN9) are taken as points that are further used for the axis determination. For the decamer this gives 40 points. I used these points to calculate the mean line (global linear helical axis) by least-squares method (searching x=x(z),y=y(z)). As a result, I got equations x=0.024*z+0.358, y=-0.073*z-1.101. The vector collinear to this line is then (-0.024,0.073,-1) but not the expected (0,0,-1) calculated by 3DNA. Therefore it looks like that I miss or misunderstood something. Any comments on how the global linear axis is calculated in 3DNA will be valuable.

Kateryna

[xiangjun]

Hi Kateryna,

You can verify the output from 3DNA by adding the two lines into the PDB structure file, and visualize it using e.g., RasMol.

Code: [Select]

HETATM 9998  XS    X X 999      -0.000  -0.000   0.000
HETATM 9999  XE    X X 999      -0.000  -0.000 -30.375The "equivalent C1' and RN9/YN1 atom pairs" are defined with reference to the SAME strand. Thus the algorithm applies to single stranded DNA/RNA as well. You can find further details in my SCHNAaP paper: 3DNA uses exactly the same procedures.

If you are really interested in an implementation, please have a look at the well-known FreeHelix/NewHelix program from Dr. Dickerson, downloadable from the NDB website.

HTH,

Xiang-Jun

[kateryna]

Hi Xiang-Jun,
Thank You very much for the reply. I have no doubt that 3DNA does correctly calculates the global linear helical axis. I've visualized the pdb file with XE, XS 3DNA axis points, and everything looked fine. I've just wanted to know the methodology that lies behind the global axis calculation. The SCHNAaP paper You have mentioned is not in the free access and unfortunately our Institute is not subscribed to the Mol. Biology. Therefore I would be very grateful if You could share somehow that paper with me (put it here or send a copy of it  by email).
I've read the Freehelix help as You suggested and looks like I've got the point. Here are some notes that might be of interest for some curious 3DNA users:
In each strand vectors connecting C1' atoms of adjacent bases and those connecting RN9/YN1 of adjacent bases are built. Then they are put to the same origin and the mean plane is calculated by least-squares method for the end points of those vectors. The normal to this plane is the vector collinear to the global linear helical axis. Hope this will help somebody.

with best regards,
Kateryna

[xiangjun]

Hi Kateryna,

Thanks for your feedback and for summarizing the issue so clearly: Other users will certainly benefit from your experience.

In the spirit of community, everyone should try to contribute as much as possible. You have set up a good example and hopefully more 3DNA users would follow your lead.

Xiang-Jun

PS. I have sent you an email with PDF attachments of my SCHNAaP/SCHNArP papers.