Indexing in for loop

I would like to create a new variable, cells_high_MI, that corresponds to the cellspikes index of the cells that have a high MISpikeTotal value. Here is my code for the MISpikeTotal calculation (towards the bottom of the code):

for i = 1:n_cells
    spikes = cellspikes{i}; % access cellspikes
    spikes = spikes./1000; % ms to secs
    spikes = spikes(spikes>startTime&spikes<stopTime);
    edgesT = linspace(startTime,stopTime,numel(trackingtimes)+1);
    binnedSpikes = histcounts(spikes,edgesT);% sorts spikes into bins
    % also bin headangle data
    n_bins_angle = 60;
    edgesHD = linspace(min(headangle), max(headangle), n_bins_angle +1);
    [occupancy,~,angle_inds] = histcounts(headangle,edgesHD);
    for iBin = 1:n_bins_angle
        spikesPerAngle(iBin) = sum(binnedSpikes(angle_inds == iBin));
    end
    SPA(1,i)={spikesPerAngle};
    % compute average firing rate for each HD angle
    firing_rate = spikesPerAngle./(occupancy.*deltaT); %this vector is a TUNING CURVE!
    FR(1,i) = {firing_rate};
    
    % now compute individual pieces needed to compute mutual info
    probability_density = occupancy./sum(occupancy); %proportion of time spent at each HD angle per total occupancy at all angles
    PD(1,i)={probability_density};
    % compute average firing rate across all HD angles
    mean_rate = numel(spikes)./(stopTime-startTime);
    MR(1,i) = mean_rate;
    % compute mutual info between firing rate and HD
    mutualInfo = sum(firing_rate .* log2(firing_rate./mean_rate) .* probability_density,'omitnan');
    MIperspike = mutualInfo./(mean_rate);
    %store mutual info for each cell into a matrix
    mutualInfoTotal(1,i) = mutualInfo; %in bits per second
    MISpikeTotal(1,i) = MIperspike;
    idx = mutualInfoTotal >= 5; 
    highMI_sec = mutualInfoTotal(idx);
end

So, I have those values of highMI_sec that contain the mutual information values for the cells with MI values of at least 5. But what I actually want are the index numbers of the cells with at least MI values of 5, because I want to use these index numbers in the following calculations:

j=1; 
for c=1:n_cells
    spikes = cellspikes{c};
    spikes = spikes(spikes > startTime & spikes < stopTime);
    if length(spikes) >= 1000
        edgesT = linspace(startTime,stopTime,numel(trackingtimes)+1);
        binnedSpikes = histcounts(spikes,edgesT);
        binnedSpikes = smoothdata(binnedSpikes,2,'gaussian',50);
        pcaBinnedSpikes(j,:) = zscore(binnedSpikes);
        j = j+1;
    end
end

So I would like to use only the cellspikes that correspond to cells with an MI of at least 5 by using the indexes of the values in mutualInfoTotal.

Thanks! :)

Answer 1

[Rafael S.T. Vieira]

Why don't you use idx as a vector?

for i = 1:n_cells
    ...
    idx(i) = mutualInfoTotal >= 5; 
    highMI_sec = mutualInfoTotal(idx(i));
end

Then, later you can skip 0 values using:

for c=1:n_cells
    if ~idx(c)
        continue
    end
    spikes = cellspikes{c};
    ...
end

[Mary Hemler]

I tried this, but I'm getting this error: Unable to perform assignment because the left and right sides have a different number of elements.

Here is the code:

for i = 1:n_cells
    spikes = cellspikes{i}; % access cellspikes
    spikes = spikes./1000; % ms to secs
    spikes = spikes(spikes>startTime&spikes<stopTime);
    edgesT = linspace(startTime,stopTime,numel(trackingtimes)+1);
    binnedSpikes = histcounts(spikes,edgesT);% sorts spikes into bins
    % also bin headangle data
    n_bins_angle = 60;
    edgesHD = linspace(min(headangle), max(headangle), n_bins_angle +1);
    [occupancy,~,angle_inds] = histcounts(headangle,edgesHD);
    for iBin = 1:n_bins_angle
        spikesPerAngle(iBin) = sum(binnedSpikes(angle_inds == iBin));
    end
    SPA(1,i)={spikesPerAngle};
    % compute average firing rate for each HD angle
    firing_rate = spikesPerAngle./(occupancy.*deltaT); %this vector is a TUNING CURVE!
    FR(1,i) = {firing_rate};
    
    % now compute individual pieces needed to compute mutual info
    probability_density = occupancy./sum(occupancy); %proportion of time spent at each HD angle per total occupancy at all angles
    PD(1,i)={probability_density};
    % compute average firing rate across all HD angles
    mean_rate = numel(spikes)./(stopTime-startTime);
    MR(1,i) = mean_rate;
    % compute mutual info between firing rate and HD
    mutualInfo = sum(firing_rate .* log2(firing_rate./mean_rate) .* probability_density,'omitnan');
    MIperspike = mutualInfo./(mean_rate);
    %store mutual info for each cell into a matrix
    mutualInfoTotal(1,i) = mutualInfo; %in bits per second
    MISpikeTotal(1,i) = MIperspike;
    idx(i) = mutualInfoTotal >= 5; 
    highMI_sec = mutualInfoTotal(idx(i));
end

[Rafael S.T. Vieira]

Sorry, I didn't see mutualInfoTotal was already a vector, which makes idx also a vector. In such case, If you want the indexes, we can find it with

pos=find(mutualInfoTotal >= 5);

which will give the indexes that you seek.