Efficient and accurate determination of the overall rotational diffusion tensor of a molecule from (15)N relaxation data using computer program ROTDIF.

We present a computer program ROTDIF for efficient determination of a complete rotational diffusion tensor of a molecule from NMR relaxation data. The derivation of the rotational diffusion tensor in the case of a fully anisotropic model is based on a six-dimensional search, which could be very time consuming, particularly if a grid search in the Euler angle space is involved. Here, we use an efficient Levenberg-Marquardt algorithm combined with Monte Carlo generation of initial guesses. The result is a dramatic, up to 50-fold improvement in the computational efficiency over the previous approaches. This method is demonstrated on a computer-generated and real protein systems. We also address the issue of sensitivity of the diffusion tensor determination from (15)N relaxation measurements to experimental errors in the relaxation rates and discuss possible artifacts from applying higher-symmetry tensor model and how to recognize them.