Structure Calculation & Analysis - Applications & Techniques in NMR Research
Scientists at Accelrys have used FELIX for the following structure calculations of NMR data.

Improved Accuracy and Precision of NMR Refined Structures Using Direct Methods

Comparison of Molecular Flexibility and Conformational Variability


-----------------------------------------------------------------------------------------------------------------------------------------

Improved Accuracy and Precision of NMR Refined Structures Using Direct Methods

99kb

Using NOE crosspeak volume buildups during refinement significantly improves the accuracy and precision of the structures generated[12]. Recent validation studies[13] have demonstrated that by refining NOE volumes directly[14], rather than derived distance restraints, the precision and the accuracy of a family of Zinc Rubredoxin structures generated from simulated data of the minimized X-ray target structure were significantly improved.

 

 

 


The average pairwise RMS to the average backbone coordinates of 30 structures generated using distance restraints only in a DG/rMD protocol was 0.59, and this RMS was reduced to 0.33 when NOE volume restraints were directly refined using an identical annealing protocol. These direct methods were also applied using the experimental data of Blake et al.[3] Figure 9 (99kb) emphasizes the improved convergence of the tyr-12 side-chain when NOE volume restraints, including internal motions[11,12], were used directly for refinement[15].

The computational efficiency of Accelrys' direct method for the refinement of NOE crosspeak volumes is directly linked to the efficiency of back-calculating NOE intensities and their gradients at nearly every timestep of a dynamics calculation. A novel algorithm termed Matrix Doubling has been developed, which dramatically improves the efficiency of the back-calculation of NOE intensities[12] without compromising accuracy. For systems containing 500 proton spins at correlation times of 3 ns and four mixing times between 50 and 200 ms, the efficiency of the calculation is approximately 100 times faster than traditional matrix diagonalization approaches. These computational improvements permit refinements that used to take days to weeks to calculate on state-of-the-art supercomputers or servers, to complete in hours to days on modern desktop workstations.


-----------------------------------------------------------------------------------------------------------------------------------------

Comparison of Molecular Flexibility and Conformational Variability
 

200kb

A common dilemma in NMR-based structure determination is by simply inspecting an ensemble of structures, it is very difficult to determine if the regions of uncertainty within the model are due to an incomplete set of restraints (due to spectral overlap or experimental artifacts) or inherent flexibility within the solution structure.

 

 

 

With the recent introduction of relaxation measurements to measure inherent flexibility[16], the spectroscopist can now readily differentiate between under-constrained regions as arising from high-mobility (since some of the NOEs tend to be conformationally averaged to zero) or too few restraints within a rigid region of the molecule. As an example, the conformational variability for the ensemble NMR model[17] generated for apo Calbindin D9k is depicted in the right-hand panel of Figure 11(200kb) as a variable width ribbon, where the diameter of the ribbon is proportional to the average heavy-atom per-residue RMS to the average coordinates after the backbone atoms have been superimposed. The ribbon is color-coded according to the hydrophobicity, or charge, of the given residue at that position along the backbone. The left-hand panel in Figure 11 (200kb) depicts a similar variable width ribbon, except that the width is varied according to the value of the generalized order parameter (inherent backbone mobility) calculated for the N-H vector from R1, R2 and NOE measurements derived from 15N-directed relaxation experiments[17].

In this example, the conformational variability of the NMR model is correlated with the regions of high mobility determined experimentally (the diameter of the ribbon is approximately the same in both images). The scientist can conclude that the unconstrained loop at the upper right is inherently mobile and no further work needs to be conducted to try and extract additional constraints in this region. The differences observed in the two calcium-binding loops pointing downwards are not significant, since the relaxation measurements were conducted on the calcium-loaded form, while the structure calculations were conducted on the apo form (no hydrogen bonds or electrostatic interactions included).


-----------------------------------------------------------------------------------------------------------------------------------------

Back to top
More Case Studies

   
     Product Overview

     Felix 2D/ND
    
Modules
        Felix 2D/ND
        Felix Assign
        Felix Autoscreen
        Felix Model
     Case Studies
        Spectral Assignment
        Spectral Processing & Analysis
        Structure Calculation & Analysis

     System Requirements
     How to Try/Buy Felix 
         
         
        www.FelixNMR.com 2007