deernet.m

From Spinach Documentation Wiki
Jump to: navigation, search

Uses pre-trained neural networks to extract distance distributions from DEER data. See our paper (https://dx.doi.org/10.1126/sciadv.aat5218) for further details.

Syntax

   dataset=deernet(deer_trace,time_axis)

Arguments

   deer_trace - experimental DEER trace, a real column vec-
                tor, must be phased into pure absorption and
                cropped on the left to make the first point
                correspond to the echo modulation maximum

   time_axis  - experimental time axis in seconds, a column
                vector; must start at zero and have uniform
                time step

Outputs

The function displays a figure if no outputs are requested, or returns a data structure with the following fields.

   dataset.input.

       time_axis  - time axis, as supplied by the user

       deer_trace - DEER trace, as supplied by the user

   dataset.deernet.

       library    - the parameters of the library with which 
                    current DEERNet had been trained

       resamp_axis  - time axis after the data was resampled
                      to match DEERNet inputdimension

       resamp_trace  - DEER trace after the data was resampl-
                       ed to match DEERNet input dimension

       deernet_rmin  - left edge of the distance grid as com-
                       puted from the user-supplied time trace
                       duration and DEERNEt input point count

       deernet_rmax  - right edge of the distance grid as com-
                       puted from the user-supplied time trace
                       duration and DEERNEt input point count

       input_rmin  - left edge of the distance grid as compu-
                     ted from the user-supplied time trace du-
                     duration and point count

       input_rmax  - right edge of the distance grid as compu-
                     ted from the user-supplied time trace du-
                     duration and point count

   dataset.output.

       background_mean - arithmetic mean of the background sig-
                         nal over the netset

       background_lb - 95% confidence interval lower bound for
                       the background signal obtained from net-
                       set statistics

       background_ub - 95% confidence interval upper bound for
                       the background signal obtained from net-
                       set statistics

       dist_distr_mean - arithmetic mean of the distance dist-
                         ributions over the netset

       dist_distr_lb - 95% confidence interval lower bound for
                       the distance distribution obtained from
                       netset statistics

       dist_distr_ub - 95% confidence interval upper bound for
                       the distance distribution obtained from
                       netset statistics

       dist_axis - the distance axis matching the distance dis-
                   ribution and satisfying the constraints im-
                   posed by both the timing of the input signal
                   or the limits of DEERNet, whichever is more
                   stringent

       mdepth - modulation depth estimate

Examples

First, the experimental data is loaded:

    % Load experimental data
    [deer_trace,time_axis]=elexsys2deernet('data_deeran/CT_DEER_mix_28_36/CT_DEER_mix_28_36');

The data is then sent to DEERNet:

    % Run deernet analysis
   deernet(deer_trace,time_axis);

With no output arguments requested in the call to DEERNet, the following plot is produced.

Deernet example.png

Further examples are available under examples/deernet/

  • example_set_paper.m - The six examples used in the paper on DEERNet.
  • example_set_deeran.m - Selection of the examples included with the DeerAnalysis package.

Notes

  1. Do not supply excessively long baseline tails in the input data - unlike Tikhonov tools, DEERNet does not need those, it works better if you cut off baseline tails.
  2. The input DEER trace must be cropped on the left to make the first point of the trace correspond to the maximum.

Deernet preprocess.png

This is done automatically in elexsys2deernet.m, and this is the only preprocessing required by DEERNet - it can handle background and noise.

See also

Neural network module

DEER/PELDOR experiments


Version 2.5, authors: Ilya Kuprov, Steve Worswick, Gunnar Jeschke