3D reconstruction of paramagnetic centre probability density from PCS data using Kuprov-Charnock equation (http://dx.doi.org/10.1039/C4CP03106G).
This function implements the three-dimensional paramagnetic centre probability density algorithm described in our forthcoming paper on the subject. The user needs to supply atomic coordinates and pseudocontact shifts. The algorithm uses Tikhonov regularisation. In the argument description below, "source" refers to the probability density of the paramagnetic centre that generates the pseudocontact shift, and "solution" refers to the pseudocontact shift itself.
parameters.xyz - nuclear coordinates as [x y z] with multiple rows at which PCS has been measured, in Angstrom. parameters.xyz_all - atomic coordinates as [x y z] with multiple rows for all atoms in the structure, in Angstrom. parameters.expt_pcs - pseudocontact shift in ppm at each nucleus. parameters.chi - effective magnetic susceptibility tensor, in units of Angstrom^3. parameters.margins - a six-element vector specifying margins to take around the bounding box of the nuclear coordina- tes supplied, to eliminate the effects of the periodic boundary conditions. parameters.box_cent - Cartesian coordinates of the centre of the source box, in Angstrom parameters.box_size - size of the source box in X, Y, and Z directions, in Angstrom parameters.plot - a cell array of strings specifying the plotting options at each iteration: 'diagnostics' - diagnostic output 'density' - probability density 'molecule' - molecular structure 'tightzoom' - zooms the plot to the molecular bounding box 'box' - source box parameters.confine - source confinement radii (a two-element vector in Angstrom). The density would only be varied in points that are no closer than the first ra- dius to any atom and no further away than the second radius from the nearest atom. parameters.sharpen - source contrast control; leave at zero unless you know what you are doing. parameters.guess - optional initial guess for the source density. If this is supplied on a different grid from the current one, it will be resampled by cubic interpolation parameters.equation - 'poisson' to recover the right hand side of the Poisson's equation, 'kuprov' to recover the probability density. parameters.gpu - set to 1 of you have an NVidia GPU, otherwise set to 0. npoints - number of grid points in each dimension of the solution cube, a positive integer; at least 64 is recommended. lambda - Tikhonov regularization parameter; start with 1, then use lcurve.m function to determine it more precisely
source_cube - source term cube with dimensions ordered as [X Y Z]. ranges - Cartesian axis extents for the source cube as [xmin xmax ymin ymax zmin zmax] in Angstroms. pred_pcs - pseudocontact shifts produced by the source cube returned in the first parameter. err_ls - least squares penalty of the final source cube reg_a - contrast penalty of the final source cube reg_b - Tikhonov penalty of the final source cube
See the pseudocontact shift analysis tutorial and the examples in nmr_paramag directory of the standard example set supplied with Spinach.
For further information on the equations and algorithms used in this function see http://dx.doi.org/10.1039/C4CP03106G