Analysis methods for populations of direction-selective ganglion cells. Accompanying a manuscript by Kühn and Gollisch.
Contact: Tim Gollisch, tim.gollisch@med.uni-goettingen.de
Website: http://www.retina.uni-goettingen.de/
Functions are written for time-binned data, here, with 30 Hz sampling rate. A minimal example for the use of each function can be found in 'MAIN.m'.
spikeCounts: activity of "N" neurons over a time of "T" time bins in N x T matrices counting spikes per time bin
stimFrames: motion steps in um per frame (time bin)
filterLen: length "L" of the filter in time bins
tbin: time bin size in seconds
nBins: number of nonlinearity bins, NLrates
NLfitType: type of applied nonlinearity function, either exponential ('exp') or exponential multiplied with a quadratic ('u-shaped')
maxBinCount: maximum number of spikes within a time bin for normalization of LN model parameters
[reconstrFrames, reconstrStimBins, filter] = getLinearPopulationReadout(spikeCounts, stimFrames, filterLen)
[totalInfo, freqBins, infoDensity, stimDensity, reconstrDensity, errorDensity] = calcMutualInformation(stimFrames, reconstrFrames, filterLen, tbin)
filter = getFilter(spikeCounts, stimFrames, filterLen)
[NLrates, NLbins] = getNL(spikeCounts, stimFrames, filter, tbin, nBins)
[PoissonSpikeCounts, fitParams] = LNmodel(stimFrames, filters, NLbins, NLrates, NLfitType, maxBinCount)
[respComp, stimComp, corrCoeff] = getCCAcomponents(spikeCounts, stimFrames, filterLen)