Improving expressions for the CSA interaction -- September 23, 2006 - 17:11

Hello Everyone,

I study structure and dynamics of RNA and have been working on a crude model free program to include expressions for the large, highly asymmetric CSA tensors (which are non-collinear with the CH/NH bond vector) diffusing in an axially symmetric highly anisotropic molecule.  I met Edward at ICMRBS '06 where I learned of the flaws of a simple grid search procedure and he encouraged me to check out the relax program.

Putting aside the daunting feeling of working with a program as huge and complex as relax, I would love to know how to modify it to accomplish what I would like to do and benefit the community as a whole, if I do it well. What I would need to do is:

- Input a table of CSA tensor values and euler orientations relating the tensor to the base/ribose frame
- read in 3 nuclei to define the base/ribose frame
- locate nearby homonuclei
- Calculate relaxation for the minimization using my expressions

For the non-RNA crowd, as I already mentioned, the CSAs of the nucleobases are very large and asymmetric.  In addition, for 13C relaxation of many nuclei, nearby homonuclear dipole interactions become significant for extended RNA molecules with tau( m ) > 5 ns.  While I focus on RNA, these modifications could easily be implemented for Proteins and DNA as well.  The expressions I presented
at ICMRBS were simply reduced for an axially symmetric prolate molecule. Really,
my expressions are nothing new and are simply revisiting some long
ignored studies by H.W. Spiess NMR Basic Principles and Progress
(1978) 15, p.55-214.  These expressions can account for a completely
asymmetric interaction diffusing
completely asymmetrically (ellipsoid?).  relax already takes
care of the extended modelfree equations. The
first 3 steps are the key ones I'd like to implement.  We assume, for
the time being, that the hetero- and homonuclear dipole mechansims and
the CSA mechanism all share the same modelfree parameters.  Unless we
measure our relaxation at many field strengths, this is a necessary
assumption to start with as there are simply way too many variables to
deal with while only having R1, R2, and NOE measurements.

What I would like to know is how to go about implementing some of these changes.  Do I need to add new functions, or edit existing ones?  Where should these edits be made?  Etc.  This is my first time participating in a large community effort and I have a lot of hopes that I/we can make this work.  Up until now, I have only coded C programs for use in our lab, but Edward assures me that coding in python is a simple switch.  I look forward to participating in the relax project.


Alex Hansen