Image align_tensor


Determine the PCS error due to structural noise via simulation.


pcs.structural_noise(align_id=None, rmsd=0.2, sim_num=1000, file=None, dir=None, force=False)

Keyword arguments

align_id: The optional alignment ID string.

rmsd: The atomic position RMSD, in Å, to randomise the spin positions with for the simulations.

sim_num: The number of simulations, N, to perform to determine the structural noise component of the PCS errors.

file: The optional name of the Grace file to plot the structural errors verses the paramagnetic centre to spin distances.

dir: The directory name to place the Grace file into.

force: A flag which if True will cause the file to be overwritten.


The analysis of the pseudo-contact shift is influenced by two significant sources of noise - that of the NMR experiment and structural noise from the 3D molecular structure used. The closer the spin to the paramagnetic centre, the greater the influence of structural noise. This distance dependence is governed by the equation:

                 sqrt(3) * abs(delta) * RMSD
    sigma_dist = --------------------------- ,

where sigma_dist is the distance component of the structural noise as a standard deviation, delta is the PCS value, RMSD is the atomic position root-mean-square deviation, and r is the paramagnetic centre to spin distance. When close to the paramagnetic centre, this error source can exceed that of the NMR experiment. The equation for the angular component of the structural noise is more complicated. The PCS error is influenced by distance, angle in the alignment frame, and the magnetic susceptibility tensor.

For the simulation the following must already be set up in the current data pipe:

The position of the paramagnetic centre.
The alignment and magnetic susceptibility tensor.

The protocol for the simulation is as follows:

The lanthanide or paramagnetic centre position will be fixed. Its motion is assumed to be on the femto- to pico- and nanosecond timescales. Hence the motion is averaged over the evolution of the PCS and can be ignored.
The positions of the nuclear spins will be randomised N times. For each simulation a random unit vector will be generated. Then a random distance along the unit vector will be generated by sampling from a Gaussian distribution centered at zero, the original spin position, with a standard deviation set to the given RMSD. Both positive and negative displacements will be used.
The PCS for the randomised position will be back calculated.
The PCS standard deviation will be calculated from the N randomised PCS values.

The standard deviation will both be stored in the spin container data structure in the relax data store as well as being added to the already present PCS error (using variance addition). This will then be used in any optimisations involving the PCS.

If the alignment ID string is not supplied, the procedure will be applied to the PCS data from all alignments.

The relax user manual (PDF), created 2016-10-28.