sc_cluster_s.m

function [c_clustid] = sc_cluster_s(s, k, varargin)
%sc_cluster_s - cluster cells using cell embeding s
%
%see also: sc_cluster_x

%if min(size(s))>3, error('S is coordinates of dimensional
%reduction.'); end

if nargin < 2, k = 6; end
p = inputParser;
defaultType = 'kmeans';
validTypes = {'kmeans', 'kmedoids', 'dbscan', ...
    'spectclust', 'snndpc', 'mbkmeans'};
%
checkType = @(x) any(validatestring(x, validTypes));

checkK = @(x) (x > 0) && isnumeric(x) && isscalar(x);

addRequired(p, 's', @isnumeric);
addRequired(p, 'k', checkK);
addOptional(p, 'type', defaultType, checkType)
addOptional(p, 'plotit', false, @islogical)
parse(p, s, k, varargin{:})
plotit = p.Results.plotit;

switch p.Results.type
    case {'spectralcluster', 'spectclust'}
        c_clustid = spectralcluster(s, k);
    case 'kmeans'
        c_clustid = kmeans(s, k);
    case 'kmedoids'
        c_clustid = kmedoids(s, k);
    case 'dbscan'
        warning('In development. Needs parameters');
    case 'snndpc'
        c_clustid = sc_snndpc(s, k);
    case 'mbkmeans'
        [~, ~, c_clustid] = pkg.mbkmeans(s, k);
end

if plotit
    gui.i_gscatter3(s, c_clustid);
    hold on
    for i = 1:k
        si = s(c_clustid == i, :);
        si = mean(si);
        if size(s, 2) == 3
            text(si(:, 1), si(:, 2), si(:, 3), sprintf('%d', i), ...
                'fontsize', 20, 'FontWeight', 'bold', 'BackgroundColor', ...
                'w', 'EdgeColor', 'k');
        else
            text(si(:, 1), si(:, 2), sprintf('%d', i), ...
                'fontsize', 20, 'FontWeight', 'bold', 'BackgroundColor', ...
                'w', 'EdgeColor', 'k');
        end
    end
end