Skip to content

Latest commit

 

History

History
81 lines (53 loc) · 3.63 KB

readme.md

File metadata and controls

81 lines (53 loc) · 3.63 KB
Raw

Bayesian decoding from spike trains

These functions perform decoding of up to two variables from neural spike train data using a Bayesian approach. It includes two main functions: SetDecParams defines options for the decoder. DecodingAnalysis then takes these options as an input, together with a structure containing the variables to decode (called Nav here) and an array of responses from which variables will be decoded (e.g Spk.spikeTrain). See the Load function section to see how the Nav and Spk structures should be constructed.

SetDecParams

The SetDecParams function allows you to define parameters required for decoding positions from spike trains. It takes as inputs a structure (Nav) containing the variables to decode and the response data (Srep). The output is a structure decparams containing options for the decoding analysis.

Parameters included in decparams:

  • subset: conditions and values used to define the subset over which the decoder will be trained.
  • cellidx: Logical array to select a subset of cells for decoding.
  • sampleRate: Sampling rate of the data.
  • scalingFactor: Scaling factor applied to the response data.
  • Xvariablename, XsmthNbins, Xbinedges: Parameters for decoding the X-axis position.
  • Yvariablename, YsmthNbins, Ybinedges: Parameters for decoding the Y-axis position.
  • occ_th: Occupancy threshold for decoding inclusion.
  • nspk_th: Minimal spikes required for decoding.
  • kfold: Number of folds for cross-validation.
  • dectimewin: Size of the decoding window.

DecodingAnalysis

The DecodingAnalysis function performs decoding of up to two variables (X and Y positions) from a set of spike trains using 2D tuning curves. Given a set of variables (stored in Nav), spike train responses (e.g. Spk.spikeTrain), and decoding parameters (decparams), this function computes various decoding results and analyses. Although the decoder may be trained on a subset of the data, decoding is performed over the entoire length of the responses. To decode over the trainin set, DecodingAnalysis uses a k-fold cross-validation approach.

Results included in the Dec structure:

  • decparams: Input parameters for decoding.
  • Xbincenters, nXbins, Ybincenters, nYbins: Binning information.
  • mapXY, mapXY_cv: Place field maps and cross-validated maps.
  • occmap, X, Y: Occupancy map and discretized position values.
  • Decoded positions and errors (XDecMax, YDecMax, XErrMax, YErrMax, etc.).
  • Confusion matrices (Xdecmat, Ydecmat) estimated over the training set.

Usage

  1. Load behavioral data and spike data:

    datapath = 'path/to/your/data'

    loadparams = SetLoadParams(datapath);

    Nav = LoaddataNav(loadparams);

    Spk = LoaddataSpk(loadparams, Nav.sampleTimes);

    Srep = Spk.spikeTrain;

  2. Define parameters using SetDecParams:

    decparams = SetDecParams(Nav, Srep);

  3. Modify paramters in decparams if needed. For instance:

    decparams.subset.Condition = [1 3 5];

    decparams.subset.Condition_op = 'ismember';

    decparams.subset.Spd = 2.5;

    decparams.subset.Spd = '>=';

    decparams.Xvariablename = 'Xpos';

    decparams.Xbinedges = 0:2:100;

    decparams.Yvariablename = [];

  4. Perform decoding using DecodingAnalysis

    Dec = DecodingAnalysis(Nav, Srep, decparams);

For more detailed information and usage examples of each function, please refer to the function documentation, Tutorial3.1 and Tutorial_handson.

Developed by J. Fournier in August 2023 for the Summer school "Advanced computational analysis for behavioral and neurophysiological recordings."