server-scripts/qhtcp-workflow/apps/matlab/easy/DMPexcel2mat.m

%% CALLED BY EASYconsole.m %%

global mpdmFile
global masterPlateFile
global drugMediaFile
global matDir

% If we already have mpdmFile, don't recreate
if (exist(mpdmFile, 'file') && ~isempty(mpdmFile))
  fprintf('The Drug Media/MasterPlate Annotation File: %s exists, skipping DMPexcel2mat.m\n', mpdmFile);
  return
end

% Do our best to find and load a relevant MasterPlate file
if ~exist(masterPlateFile, 'file') || isempty(masterPlateFile)
  if exist(fullfile(matDir), 'dir')
    % Try to find the masterPlateFile automatically (newest created first)
    try
      files=dir(matDir);
      mpFiles={files(strncmp(files.name, 'MasterPlate_', 12)).name};
      if isempty(mpFiles)
        throw(MException('MATLAB:dir', 'No MasterPlate_ files in the default MasterPlate directory'));
      end
      % this sorts by date (newest first)
      [~, sortedIndices]=sort(datenum({files(strncmp(mpFiles.name, 'MasterPlate_', 12)).date}), 'descend');
      sortedFiles=mpFiles(sortedIndices);
      masterPlateFile=sortedFiles{1};
      disp('Using newest MasterPlate file: ', masterPlateFile, ', skipping directory selection');
    catch ME
      % This can be silent, not really an error
    end


    try
      % For standalone mode
      % GUI input for selecting a MasterPlate Excel file
      questdlg('Select MasterPlate Directory (containing DrugMedia_ & MasterPlate_ files)','Directory Selection','OK',...
      struct('Default','OK','Interpreter','tex'))
      dirToScan=uigetdir();
      files=dir(dirToScan);
      mpFiles={files(strncmp(files.name, 'MasterPlate_', 12)).name};
      if isempty(mpFiles)
        throw(MException('MATLAB:dir', 'No MasterPlate_ files in directory'))
      end
      % this sorts by date (newest first)
      [~, sortedIndices]=sort(datenum({files(strncmp(mpFiles.name, 'MasterPlate_', 12)).date}), 'descend');
      sortedFiles=mpFiles{sortedIndices};
      masterPlateFile=sortedFiles{1};
      disp('Using newest MasterPlate file: ', masterPlateFile, ', skipping directory selection');
      break
    catch ME
      h=msgbox(ME.message, 'Error', 'error');
      disp('Rerunning directory selection');
      % I don't know what else we'll need to do here
    end
  else
    try
      % For standalone mode
      % GUI input for selecting a MasterPlate Excel file
      questdlg('Select MasterPlate Directory (containing DrugMedia_ & MasterPlate_ files)','Directory Selection','OK',...
      struct('Default','OK','Interpreter','tex'))
      dirToScan=uigetdir();
      files=dir(dirToScan);
      mpFiles={files(strncmp(files.name, 'MasterPlate_', 12)).name};
      if isempty(mpFiles)
        throw (MException('MATLAB:dir', 'No MasterPlate_ files in directory'));
      end
      % this sorts by date (newest first)
      [~, sortedIndices]=sort(datenum({files(strncmp(mpFiles.name, 'MasterPlate_', 12)).date}), 'descend');
      sortedFiles=mpFiles{sortedIndices};
      masterPlateFile=sortedFiles{1};
      disp('Using newest MasterPlate file: ', masterPlateFile, ', skipping directory selection');
      break
    catch ME
      h=msgbox(ME.message, 'Error', 'error');
      uiwait(h);
      disp('Rerunning directory selection');
      % I don't know what else we'll need to do here
    end
  end
else
  fprintf('Using MasterPlate file: %s skipping directory selection\n', masterPlateFile);
end

% fid=fopen(masterPlateFile)%('exp23PrintTimes.xls'); % textread puts date and time sequentially into vector
% TODO needs explanation, it isn't clear what it is for
% The input files should be csv, not xlsx
% This whole file could be much better and more portable
if ispc
  [num, txt, raw]=xlsread(masterPlateFile);  %,'Yor1HitsMPsetFinal');
  fields={txt(2,1:15)}; %or 1:17 for later but dont wish to exceed and cause error ? if used  
else
  clear MPtbl
  opts=detectImportOptions(masterPlateFile);
  MPtbl=readtable(masterPlateFile,opts);
  MPtbl=readtable(masterPlateFile);
  fields={opts.VariableNames};  %? if used anywhere although 'saved' to MPDMmat
  MPcell=readcell(masterPlateFile);
end

numb=0;
clear MP;
try
  if ispc
    excLnNum=3;
    while (isequal(txt{excLnNum,1},'###'))
      numb=numb+1;    
      MP(numb).head={raw(excLnNum,2:6)};
      MP(numb).recNum={raw((excLnNum+1):(excLnNum+384),1)};
      MP(numb).orf={raw((excLnNum+1):(excLnNum+384),2)};
      MP(numb).strain={raw((excLnNum+1):(excLnNum+384),3)};
      MP(numb).genename={raw((excLnNum+1):(excLnNum+384),12)};
      MP(numb).drug={raw((excLnNum+1):(excLnNum+384),8)};
      MP(numb).media={raw((excLnNum+1):(excLnNum+384),7)};
      if size(raw,2)>15
        MP(numb).orfRep={raw((excLnNum+1):(excLnNum+384),16)}; % added 12_1005 to specify replicates Orfs in MP
        MP(numb).specifics={raw((excLnNum+1):(excLnNum+384),17)}; % added 12_1008 to specify replicates Specific details in MP
      else
        MP(numb).orfRep=' ';   
        MP(numb).specifics= ' ';
      end
      % Future MP field
      % if size(raw,2)>17 
      %   MP(numb).specifics2= {raw((excLnNum+1):(excLnNum+384),18)};  % added 12_1008 to specify strain Bkground in MP
      % else
      %   MP(numb).specifics2=' ';
      % end
      excLnNum=excLnNum+385;
      msg=strcat('NumberOfMP=',num2str(numb), '   lastLineNo.=',num2str(excLnNum));
    end
  else 
    excLnNum=1;
    while (isequal(MPcell{(excLnNum+2),1},'###'))
      numb=numb+1;    
      MP(numb).head={MPtbl(excLnNum,2:6)};
      MP(numb).head{1}=table2cell(MP(numb).head{1});
      MP(numb).recNum={MPtbl((excLnNum+1):(excLnNum+384),1)};
      MP(numb).recNum{1}=table2cell(MP(numb).recNum{1});
      MP(numb).orf={MPtbl((excLnNum+1):(excLnNum+384),2)};
      MP(numb).orf{1}=table2cell(MP(numb).orf{1});
      MP(numb).strain={MPtbl((excLnNum+1):(excLnNum+384),3)};
      MP(numb).strain{1}=table2cell(MP(numb).strain{1});
      MP(numb).genename={MPtbl((excLnNum+1):(excLnNum+384),12)};
      MP(numb).genename{1}= table2cell(MP(numb).genename{1});
      MP(numb).media= {MPtbl((excLnNum+1):(excLnNum+384),7)};
      MP(numb).media{1}= table2cell(MP(numb).media{1});
      if size(MPtbl,2)>15
        MP(numb).orfRep= {MPtbl((excLnNum+1):(excLnNum+384),16)}; % added 12_1005 to specify replicates Orfs in MP
        MP(numb).orfRep{1}=table2cell(MP(numb).orfRep{1});
        MP(numb).specifics={MPtbl((excLnNum+1):(excLnNum+384),17)}; % added 12_1008 to specify replicates Specific details in MP
        MP(numb).specifics{1}=table2cell(MP(numb).specifics{1});
      else
        MP(numb).orfRep= ' ';   
        MP(numb).specifics= ' ';
      end
      excLnNum=excLnNum+385;
      msg=strcat('NumberOfMP=',num2str(numb), 'lastLineNo.=',num2str(excLnNum));
    end
  end
catch ME
  h=msgbox(ME.message, 'Error', 'error');
  uiwait(h);
end

%DMupload
%Drug and Media Plate setup  Upload from Excel
excLnNum=1;
numOfDrugs=0;
numOfMedias=0;

% Grabbing the bare filename from the MasterPlate file to see if we can automatically
% find a matching DrugMedia file
[mpFile, mpPath]=fullfile(masterPlateFile);
mpFileParts=strsplit(mpFile, '_');
mpBareFileName=strjoin(mpFileParts(2:end-1), '_');

if ~exist(drugMediaFile, 'file') || isempty(drugMediaFile)
  if exist(fullfile(matDir), 'dir')
    try
      dmFileToTest=fullfile(mpPath, 'DrugMedia_', mpBareFileName, '.xlsx');
      if exist(dmFileToTest, 'file') % Try to find a matching drug media file
        drugMediaFile=dmFileToTest;
        fprintf('Using matching DrugMedia file: %s, skipping directory selection\n', drugMediaFile);
      else
        % Try to find the DrugMedia file automatically (newest created first)
        files=dir(matDir);
        dmFiles={files(strncmp(files.name, 'DrugMedia_', 10)).name};
        if isempty(dmFiles)
          throw (MException('MATLAB:dir', 'No DrugMedia_ files in directory'));
        end
        % this sorts by date (newest first)
        [~, sortedIndices]=sort(datenum({files(strncmp(dmFiles.name, 'DrugMedia_', 10)).date}), 'descend');
        sortedFiles=dmFiles{sortedIndices};
        drugMediaFile=sortedFiles{1};
        fprintf('Using newest DrugMedia file: %s, skipping directory selection\n', drugMediaFile);
      end
    catch Me
      % This can be silent, not really an error
    end
  else
    for i=1:5 % give users 5 chances to get it right
      try
        % GUI input for selecting a DrugMedia file
        % sort by newest matching DrugMedia file and return that if we find one
        % For standalone mode
        % GUI input for selecting a MasterPlate Excel file
        questdlg('Select DrugMedia directory','Directory Selection','OK',...
        struct('Default','OK','Interpreter','tex'))
        dirToScan=uigetdir();
        files=dir(dirToScan);
        dmFiles={files(strncmp(files.name, 'DrugMedia_', 10)).name};
        if isempty(dmFiles)
          throw (MException('MATLAB:dir', 'No DrugMedia_ files in directory'));
        end
        % this sorts by date (newest first)
        [~, sortedIndices]=sort(datenum({files(strncmp(dmFiles.name, 'DrugMedia_', 10)).date}), 'descend');
        sortedFiles=dmFiles{sortedIndices};
        drugMediaFile=sortedFiles{1};
      catch ME
        h=msgbox(ME.message, 'Error', 'error');
        uiwait(h);
        disp('Rerunning directory selection');
        % I don't know what else we'll need to do here
      end
    end
  end
else
  fprintf('Using drugMediaFile: %s, skipping directory selection\n', drugMediaFile);
end

% Drug and Media Plate setup
% TODO needs better explanation
if ispc
  [num, txt, raw]=xlsread(drugMediaFile);  %'Yor1HitsMPsetFinal');
  fields={txt(2,1:5)};
  Linked=num(1,1);
else
  opts=detectImportOptions(drugMediaFile); 
  DMtbl=readtable(drugMediaFile,opts);
  fields=opts.VariableOptions;   
  Linked=DMtbl{1,1};
  DMcell=readcell(drugMediaFile);
end

% TODO needs better explanation
numb=0;
if isequal(Linked,1) % Drugs and Media are linked 1 to 1; else they are combinatorial
  clear DM;
  excLnNum=2;
  if ispc
    while (~isequal(txt{excLnNum,2},'###'))
      numb=numb+1;    
      DM.drug(numb)={raw(excLnNum,2)};
      DM.conc(numb)={raw(excLnNum,3)};
      DM.media(numb)={raw(excLnNum,4)};
      DM.mod1(numb)={raw(excLnNum,5)};
      DM.conc1(numb)={raw(excLnNum,6)};
      DM.mod2(numb)={raw(excLnNum,7)};
      DM.conc2(numb)={raw(excLnNum,8)};
      excLnNum=excLnNum+1;
      msg=strcat('NumberOf1:1DrugMediaPlates=',num2str(numb), '   lastLineNo.=',num2str(excLnNum));
    end
  else
    clear DM
    excLnNum=1;
    while (~isequal(DMcell{excLnNum+1,2},'###'))
      numb=numb+1;    
      DM.drug(numb)={DMtbl(excLnNum,2)};
      DM.drug(numb)=table2cell(DM.drug{numb});
      DM.conc(numb)={DMtbl(excLnNum,3)};
      DM.conc(numb)=table2cell(DM.conc{numb});
      DM.media(numb)={DMtbl(excLnNum,4)};
      DM.media(numb)=table2cell(DM.media{numb});
      DM.mod1(numb)={DMtbl(excLnNum,5)};
      DM.mod1(numb)=table2cell(DM.mod1{numb});
      DM.conc1(numb)={DMtbl(excLnNum,6)};
      DM.conc1(numb)=table2cell(DM.conc1{numb});
      DM.mod2(numb)={DMtbl(excLnNum,7)};
      DM.mod2(numb)=table2cell(DM.mod2{numb});
      DM.conc2(numb)={DMtbl(excLnNum,8)};
      DM.conc2(numb)=table2cell(DM.conc2{numb});
      excLnNum=excLnNum+1;
      msg=strcat('NumberOf1:1DrugMediaPlates=',num2str(numb), '   lastLineNo.=',num2str(excLnNum));
    end
  end
end
% Legacy contengency: not ever used
if isequal(Linked,0) % 0 indicates Drugs and Media are combinatorial
  clear DM;
  excLnNum=2;
  drgCnt=0;
  medCnt=0;
  if ispc
    while (~isequal(txt{excLnNum,2},'###'))
      drgCnt=drgCnt+1;    
      DM.drug(drgCnt)={raw(excLnNum,2)};
      DM.conc(drgCnt)={raw(excLnNum,3)};
      excLnNum=excLnNum+1;
    end

    while (~isequal(txt{excLnNum,4},'###'))
      medCnt=medCnt+1;    
      DM.media(medCnt)={raw(excLnNum,4)};
      excLnNum=excLnNum+1;
    end
  else
    excLnNum=1;
    while (~isequal(DMcell{excLnNum+1,2},'###'))
      drgCnt=drgCnt+1;    
      DM.drug(drgCnt)={DMtbl(excLnNum,2)};
      DM.conc(drgCnt)={DMtbl(excLnNum,3)};
      excLnNum=excLnNum+1;
    end

    while (~isequal(DMcel{excLnNum+1,4},'###'))
      medCnt=medCnt+1;    
      DM.media(medCnt)={DMtbl(excLnNum,4)};
      excLnNum=excLnNum+1;
    end
  end
  msg=strcat('NumberOfDrugs=',num2str(drgCnt), '     NumberOfMedias=',num2str(medCnt)  );
end

save(mpdmFile, 'fields','MP','DM','Linked');
msgbox(sprintf('Drug-Media-MasterPlate Annotation File %s Generation Complete', mpdmFile))