clear all; clc

folder = 'ALL MOD DATA'; %// Place directory here

%// Get all MAT files in directory
f = dir(['' folder '\**\*.txt']);

%Add path
addpath(genpath(folder));
S = length(f);

answer = inputdlg({'Average RT (min)',...
    'RT tolerance (min)',...
    'MZ1',...
    'MZ2',...
    'MZ3'},...
    'Choose please',1,{'22.56','0.07','57','43','71'})'; % Defaults line
if isempty(answer),return;end
%
% answer = inputdlg({'Average RT (min)',...
%     'RT tolerance (min)',...
%     'MZ1',...
%     'MZ2',...
%     'MZ3',...
%     'MZ4'},...
%     'Choose please',1,{'2.49','0.1','31','45','46','29'})'; % Defaults line
% if isempty(answer),return;end

Integration=zeros(400,S); %400 (rows) is number of peaks per file
N_Peaks=zeros(S,1);
RT=zeros(400,S);MZ=zeros(400,5,S);
for f_ind=1:S
    fid=fopen(f(f_ind).name);
    A=fscanf(fid,'%c');
    fclose(fid);
    %% RT & mz
    a=regexp(A,'PEAK LIST OVERVIEW','once');
    b=regexp(A,'PEAK INTEGRATION RESULTS','once');
    tmp=regexp(A(a:b),'\[\d+\][^\r]+','match');
    for k=1:length(tmp)
        RT_tmp=regexp(tmp{k},'\d+\.\d+','match','once');
        RT(k,f_ind)=str2num(RT_tmp); %#ok
        mz_tmp=regexp(tmp{k},'[\d\.\s]+\.\.\.','match','once');
        mz_tmp2=str2num(mz_tmp(1:end-3)); %#ok
        MZ(k,:,f_ind)=0;
        MZ(k,1:length(mz_tmp2),f_ind)=mz_tmp2;
    end
    
    %% Find the peaks
    a=regexp(A,'PEAK INTEGRATION RESULTS','once');
    b=regexp(A,'SCAN IDENTIFICATION RESULTS','once');
    tmp=regexp(A(a:b),'\d+\.\d+','match');
    I_tmp=str2double (tmp);
    Integration(1:length(I_tmp),f_ind)=I_tmp;
    N_Peaks(f_ind)=length(I_tmp);
end

%% Peak Hunter
%read from input
RT_avg = str2double(answer{1});
Tol = str2double(answer{2});
MZ1 = str2double(answer{3});
MZ2 = str2double(answer{4});
MZ3 = str2double(answer{5});
%MZ4 = str2double(answer{6});
AUC = zeros(5,S);
for f_ind=1:S
    %find RT+/-Tol range
    k = find(RT(:,f_ind)>(RT_avg-Tol) & RT(:,f_ind)<RT_avg+Tol);
    count_MZ = zeros(1,length(k));
    for k_ind=1:length(k) %for each match in RT
        %Check MZ match
        if ismember(MZ1,MZ(k(k_ind),:,f_ind))
            count_MZ(k_ind) = count_MZ(k_ind)+1;
        end
        if ismember(MZ2,MZ(k(k_ind),:,f_ind))
            count_MZ(k_ind) = count_MZ(k_ind)+1;
        end
        if ismember(MZ3,MZ(k(k_ind),:,f_ind))
            count_MZ(k_ind) = count_MZ(k_ind)+1;
        end
%         if ismember(MZ4,MZ(k(k_ind),:,f_ind))
%             count_MZ(k_ind) = count_MZ(k_ind)+1;
%         end
    end
    MZ_match_ind = find(count_MZ==3); % or 4
    %Add matched MZ to array
    if length(MZ_match_ind)==1
        AUC(1,f_ind) = Integration(k(MZ_match_ind),f_ind)/1e6;
    elseif isempty(MZ_match_ind)
        AUC(1,f_ind) = NaN;
    else
        AUC(1,f_ind) = sum(Integration(k(MZ_match_ind),f_ind))/1e6;
        AUC_temp = Integration(k(MZ_match_ind),f_ind)/1e6;
        AUC(2,f_ind) = AUC_temp(1);
        AUC(3,f_ind) = AUC_temp(2);
            if length(MZ_match_ind)==3
        AUC(4,f_ind) = AUC_temp(3);
            end
            if length(MZ_match_ind)==4
        AUC(5,f_ind) = AUC_temp(4);
            end
    end
end

%% headers
for i=1:S
    Headers{i} = {f(i).name};
end