CodonOptOutputCompiler:MATLAB

CodonOptOutputCompiler.m

function CodonOptOutputCompiler(xlsFile, outPrefix)
if nargin < 2,
    fprintf('USAGE:\n');
    fprintf('CodonOptOutputCompiler <input xls> <output prefix>\n');
    fprintf('\tinput xls\tMS Excel spreadsheet output from Codon Optimizer\n');
    fprintf('\n');
    fprintf('\toutput prefix\tPrefix for output files.\n');
    fprintf('\t             \tOutput is two files:\n');
    fprintf('\t             \t\t<output prefix>.xls\n');
    fprintf('\t             \t\t<output prefix>.fasta\n');
    
    return;
end

[status, sheets] = xlsfinfo(xlsFile);

if ~isempty(status),
    resNames = sheets(~cellfun(@isempty, regexpi(sheets, '^>')));
    
    % preallocate the results storage array
    out = cell(length(resNames), 2);
    
    % get sheets that have output from the optimizer
    % they start with '>'
    for i = 1:length(resNames),
        resName = resNames{i};
        
        % get the sequence name
        id = regexpi(resName, '(?<id>[\w\s]+?)-DNA', 'names');
        id = id.id;
        
        % get the sequence
        [~,seq,~] = xlsread(xlsFile, resName, 'B6');
        
        % write the id and sequence to a compiled list
        out(i, :) = {id, seq{1}};
        
        disp(out(i, :));
    end
    
    % write the compiled list to an xl spreadsheet
    [status, message] = xlswrite([outPrefix '.xls'], out, 'Sheet 1');
    
    if ~status,
        disp(message);
    end
    
    % write the compiled list to a fasta format text file
    fastaData = cell2struct(out, {'Header', 'Sequence'}, 2);
    fastawrite([outPrefix, '.fasta'], fastaData);
    
else
    % there was an error
    disp(sheets)
end