Determining intercalator roll compared to DNA local helical axis?

[vanarbulax]

Dear Xiang-Jun and other 3DNA users,

I’m studying acridines intercalating between DNA base-pairs using MD and would like to get a value of the intercalator “roll angle” with respect to the DNA base-planes and would appreciate some guidance as to whether my approach is valid.

Can I use the local helical axis vectors produced in 3DNA’s output? I was thinking I take the axis vector corresponding to the base-pair step my drug is intercalated in, and then take cos-1 of the dot product of that axis vector with the intercalator's short-axis to measure the roll angle. Then I can find its difference to the Roll angle given in the .out file.

I would like to compare these simulation values with experiment, we don’t have any atomic resolution data but do have electric dichroism measurements which align the DNA to an electric field and then use UV to polarise the intercalator’s short-axis and hence allow a determination of an average roll angle with respect to the base base-pairs.

To get the intercalator's short-axis vector I’m currently using VMD’s ‘orient’ plug-in to do a principal axis fit. I'm writing a .tcl so the analysis can be done automatically through VMD.

Gratefully,

Keiran

[xiangjun]

Hi Keiran,

Why not just try your ideas to see if you can get sensible results? Since you are using .tcl scripts to automate the MD analyses, you can always play around with different options.

Here is some info from the 3DNA side that you may find helpful:

• The "roll" angle is a component (the other one is "tilt") of the so-called bending angle between two base pairs (bps). In 3DNA, the bending angle is calculated from the normal vectors of the bps. Since "tilt" is normally smaller than "roll", the magnitude of "roll" is approximately equal to the bending angle.
• The "local helical axis" is defined based on the reference frames of the two bps. It is quite sensitive to local distortions. Try to draw the local helical axes of the classic Dickerson-Drew B-DNA dodecamer, you will see that they do not align well but show large variations. Please make sure that any conclusions you draw based on such "local helical axis" are robust.

See the "tech_details.pdf" distributed with 3DNA for details on how "local helical axis" is defined, and how "roll" is calculated.

Since acridine is aromatic, you could fit a least-square plane on it. The normal vector of the ls-plane should be more resistant to noises from MD simulations. You can then calculate angles between this normal vector with those of the two bps.

HTH,

Xiang-Jun