MainAnalysis_Example.m

%% User Input
%% Path information
DataDir = {'H:\MatchingUnits\RawData'}; % Raw data folders, typically servers were e.g. *.cbin files are stored
SaveDir = 'H:\MatchingUnits\Output\Concatenated1Day' % Folder where to store the results
tmpdatafolder = 'H:\MatchingUnits\Tmp'; % temporary folder for temporary decompression of data 
KilosortDir = 'H:\MatchingUnits\KilosortOutput'; % Kilosort output folder
GithubDir = 'C:\Users\EnnyB\Documents\GitHub'; % Github directory
PythonEXE = 'C:\Users\EnnyB\anaconda3\envs\pyks2\pythonw.exe' % Python version to run python code in:

%% Information on experiments
MiceOpt = {'AL032','AV008','CB016','EB019','JF067'}; % Add all mice you want to analyze
DataDir2Use = repmat(1,[1,length(MiceOpt)]); % In case you have multiple DataDir, index which directory is used for each mouse
RecordingType = repmat({'Acute'},1,length(MiceOpt)); % And whether recordings were acute (default)
RecordingType(ismember(MiceOpt,{'AL032','EB019','CB016','AV008','JF067'}))={'Chronic'}; %EB014', % Or maybe Chronic?

%% Parameters on how to prepare units/data for analysis
PrepareClusInfoparams.RunPyKSChronicStitched = 1; % Default 0. if 1, run PyKS chronic recordings stitched when same IMRO table was used
PrepareClusInfoparams.CopyToTmpFirst = 1; % If 1, copy data to local first, don't run from server (= advised!)
PrepareClusInfoparams.DecompressLocal = 1; % If 1, uncompress data first if it's currently compressed (= necessary for unitmatch and faster for QualityMetrics)

% Storing preprocessed data?
PrepareClusInfoparams.ReLoadAlways = 1; % If 1, SP & Clusinfo are always loaded from KS output
PrepareClusInfoparams.saveSp = 1; % Save SP struct for easy loading of preprocessed data
PrepareClusInfoparams.binsz = 0.01; %Bin size for PSTHs in seconds

% Quality Metrics
PrepareClusInfoparams.RunQualityMetrics = 1; % If 1, Run the quality metrics (Bombcell @JulieFabre)
PrepareClusInfoparams.RedoQM = 0; %if 1, redo quality metrics if it already exists
PrepareClusInfoparams.InspectQualityMetrics = 0; % If 1, Inspect the quality matrix/data set using the GUI (manual inspection)
PrepareClusInfoparams.loadPCs = 0; % Only necessary when computiong isoluation metrics/drift in QM. You save a lot of time keeping this at 0

% UnitMatch
PrepareClusInfoparams.UnitMatch = 1; % If 1, find identical units across sessions or oversplits in a fast and flexible way
PrepareClusInfoparams.RedoUnitMatch = 1; % if 1, Redo unitmatch
PrepareClusInfoparams.separateIMRO = 0; % Run for every IMRO separately (for memory reasons or when having multiple probes this might be a good idea)

% UnitMatch Parameters:
% All parameters to choose from: {'AmplitudeSim','spatialdecaySim','WavformMSE','WVCorr','CentroidDist','CentroidVar','CentroidDistRecentered','TrajAngleSim','TrajDistSim'};
% WavformSim is average of WVCorr and WavformMSE
% CentroidOverlord is average of CentroidDistRecentered and CentroidVar
PrepareClusInfoparams.Scores2Include = {'CentroidDist','WavformSim','CentroidOverlord','spatialdecaySim','AmplitudeSim','TrajAngleSim'}; %{'AmplitudeSim','spatialdecayfitSim','WavformSim','CentroidDist','CentroidVar','TrajAngleSim'}; % 
PrepareClusInfoparams.ApplyExistingBayesModel = 0; %If 1, use probability distributions made available by us - 
PrepareClusInfoparams.MakePlotsOfPairs = 0; % Plots pairs for inspection (UnitMatch)
PrepareClusInfoparams.AssignUniqueID = 1; % Assign UniqueID 
PrepareClusInfoparams.GoodUnitsOnly = 1; % Include only good untis in the UnitMatch analysis - faster and more sensical

%% Automatic from here
PrepareClusInfoparams.SaveDir = SaveDir; % Save results here
PrepareClusInfoparams.tmpdatafolder = tmpdatafolder; % use this as a local directory (should be large enough to handle all sessions you want to combine)

%% All dependencies you want to add (you may need to download these, all available via github)
addpath(genpath(cd))

% Required (for using UnitMatch):
addpath(genpath(fullfile(GithubDir,'spikes')))% Should work with normal spikes toolbox, but I use the forked version in https://github.com/EnnyvanBeest/spikes
addpath(genpath(fullfile(GithubDir,'npy-matlab'))) % https://github.com/kwikteam/npy-matlab
addpath(genpath(fullfile(GithubDir,'mtscomp'))) % https://github.com/int-brain-lab/mtscomp

% Advised (quality metrics for unit selection):
addpath(genpath(fullfile(GithubDir,'bombcell'))) % DOI: 10.5281/zenodo.8172822, https://github.com/Julie-Fabre/bombcell 

% Optional for histology:
addpath(genpath(fullfile(GithubDir,'AP_histology'))) % https://github.com/petersaj/AP_histology
addpath(genpath(fullfile(GithubDir,'allenCCF'))) % https://github.com/cortex-lab/allenCCF

% UNITMATCH - Move to top of paths 
addpath(genpath(fullfile(GithubDir,'UnitMatch'))) % Make sure to have this one fresh in the path (so run this last)

try
    % Python version to run python code in:
    pyversion(PythonEXE) %Explanation on how to do this is provided in the README
catch ME
    disp(ME)
end

%% Actual pipeline
%% PyKS - run pykilosort from Matlab/Python integration
RunPyKS2_FromMatlab

%% Runs unitmatch across all data from a mouse to generate a table
RunUnitMatchAllDataPerMouse

%% Across Mice Graphs
SummarizeAcrossMice