postprocess_localsensitivity.m

clear all

% This script can be used to postprocess the local sensitivity analysis
% results

fname = "./Sensitivity/2024-01-15_LocalAnalysis_50days.csv";
dat = readtable(fname);
fname = "./Sensitivity/2023-11-30_LocalAnalysis_50days_base_vals.csv";
base_vals = readtable(fname);
base_Kplas = base_vals.K_plasma;
base_Kmusc = base_vals.K_musc;

%% Get nice parameter names
ParNames = [];
for ii = 1:length(dat.ParameterName)
    pname = dat.ParameterName{ii};
    %disp(pname)
    temp = change_parname(pname);
    ParNames = [ParNames; temp];
end

%% Get percent change
up_per = zeros(length(dat.ParameterName), 2);
down_per = zeros(size(up_per));
up_down_per = zeros(size(up_per));

% increase by 10%
up_per(:,1) = (dat.Kplas_Increase - base_Kplas)./base_Kplas * 100.0;
up_per(:,2) = (dat.Kmusc_Increase - base_Kmusc)./base_Kmusc * 100.0;
% decrease by 10%
down_per(:,1) = (dat.Kplas_Decrease - base_Kplas)./base_Kplas * 100.0;
down_per(:,2) = (dat.Kmusc_Decrease - base_Kmusc)./base_Kmusc * 100.0;
% increase by 10% - decrease by 10%
up_down_per(:,1) = (dat.Kplas_Increase - dat.Kplas_Decrease)./base_Kplas * 100.0;
up_down_per(:,2) = (dat.Kmusc_Increase - dat.Kmusc_Decrease)./base_Kmusc * 100.0;

%% Make figures
xlabels = ParNames; %dat.ParameterName;
ylabels_updown = {'\Delta K_{plasma, i}', '\Delta K_{intracellular, i}'};
ylabels_up = {'\Delta K_{plasma, i}^{+10%}', '\Delta K_{intracellular, i}^{+10%}'};
ylabels_down = {'\Delta K_{plasma, i}^{-10%}', '\Delta K_{intracellular, i}^{-10%}'};
%% Figure with up 10 percent
% Up percent
% Find values less than 2%
up_per_h = round(up_per, 2,"significant"); % up_per for heatmap
[r,c] = find(abs(up_per_h) < 1.0);
for ii = 1:length(r)
    up_per_h(r(ii), c(ii)) = NaN;
end
figure(1)
clf;
h = heatmap(xlabels, ylabels_up, up_per',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
% h = heatmap(xlabels, ylabels_up, up_per_h',...
%                 'colormap', parula, ...
%                 'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
h.FontSize = 14;

%% Down 10 percent

% Down percent
down_per_h = round(down_per, 2, "significant");
[r,c] = find(abs(down_per_h) < 1.0);
for ii = 1:length(r)
    down_per_h(r(ii),c(ii)) = NaN;
end

figure(2)
clf;
h = heatmap(xlabels, ylabels_down, down_per',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
% h = heatmap(xlabels, ylabels, down_per_h',...
%                 'colormap', parula, ...
%                 'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
h.FontSize = 14;

%% Up/Down percent
% Up/Down percent
up_down_per_h = round(up_down_per, 2, "significant");
[r,c] = find(abs(up_down_per_h) < 2.0);
for ii = 1:length(r)
    up_down_per_h(r(ii), c(ii)) = NaN;
end
figure(3)
clf;
h = heatmap(xlabels, ylabels_updown, up_down_per_h',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
h.FontSize = 14;

fprintf('done \n')

%% Figures with removed values
% Remove where both NaN
rows_notAllNan = find(~all(isnan(up_down_per_h), 2));
% remove V
mask = ~ismember(rows_notAllNan, [2,4]); 
rows_notAllNan = rows_notAllNan(mask);

xlabels = ParNames(rows_notAllNan);
up_down_per_h2 = up_down_per_h(rows_notAllNan, :);
up_per_h2 = up_per_h(rows_notAllNan, :);
down_per_h2 = down_per_h(rows_notAllNan, :);

% Switch the order by largest impact on K_IC
[~, sortIDs] = sort(up_down_per_h2(:,2)); 

% shift so mKALDO is in the right index
index = 7;
% Remove the last element from the array
lastElement = sortIDs(end);
sortIDs(end) = [];
% Insert the last element at the desired index
sortIDs = [sortIDs(1:index-1); lastElement; sortIDs(index:end)];


up_down_per_h2 = up_down_per_h2(sortIDs,:);
up_per_h2 = up_per_h2(sortIDs,:);
down_per_h2 = down_per_h2(sortIDs,:);
xlabels = xlabels(sortIDs);


figure(4)
clf

minValue = min(min([up_per_h2(:); down_per_h2(:)]));
maxValue = max(max([up_per_h2(:); down_per_h2(:)]));

subplot(2,1,1)
h1 = heatmap(xlabels, ylabels_up, up_per_h2',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<1%');
h1.FontSize = 18;
% Apply the common color limits
h1.ColorLimits = [minValue, maxValue];


% figure(5)
% clf
subplot(2,1,2)
h2 = heatmap(xlabels, ylabels_down, down_per_h2',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<1%');
h2.FontSize = 18;
h2.ColorLimits = [minValue, maxValue];



%% Up down per

figure(6)
clf;
h = heatmap(xlabels, ylabels_updown, up_down_per_h2',...
                'colormap', parula, ...
                'MissingDataColor', 'w', 'MissingDataLabel', '<2%');
h.FontSize = 14;

fprintf('done \n')

% Up_per with rows_notAllNan