Homec4science
Diffusion Simon Botrylloides Regeneration Group (master) Browse

SiBoRG/helpers/parse_strains_db.m
24e659740c26master

Last Change
Raw File

parse_strains_db.m
View Options

% PARSE_STRAINS_DB extracts from a SciNote export CSV file the clones'
% history and relations.
%
% STRAINS = PARSE_STRAINS_DB(CSV_FILE) extracts STRAINS from CSV_FILE.
% STRAINS is a matrix with the following format:
% [ StrainID CloneID ParentSlide SlideID CreationDate DeathDate ]
%
% Blanchoud Group, UNIFR
% Simon Blanchoud
% 01/12/2020
function strains = parse_strains_db(fname)
% Prepare the output
strains = NaN(0,6);
% Only process files
if (isfile(fname))
fid = fopen(fname, 'r');
% Make sure we can read it
if fid >= 0
% Process the header line
line = fgetl(fid);
headers = strsplit(line, ',');
% Prepare a cell matrix to store the CSV file
data = cell(0, length(headers));
% Process the CSV file
line = fgetl(fid);
while ischar(line)
% In case there are \", then there is likely a problematic comma
if (any(line == '"'))
ids = find(line == '"');
ids = reshape(ids, 2, []);
% We simply replace the commas with \;
for i=1:size(ids, 2)
tmpstr = line(ids(1,i):ids(2,i));
tmpstr(tmpstr==',') = ';';
line(ids(1,i):ids(2,i)) = tmpstr;
end
% And we delete the \"
line(line == '"') = [];
end
% We simplify a bit the symbols used
line(line=='.') = '/';
line(line==':') = ';';
% And we store the line of the CSV into a cell matrix
data(end+1,:) = strsplit(line, ',', 'COLLAPSEDELIMITERS', false);
% Continue the file
line = fgetl(fid);
end
end
% Assign the variables
nclones = size(data,1);
nchars = 1;
strains = NaN(nclones, 6);
% Loop through the CSV table
for i=1:nclones
% Make sure this is a proper strain data
nstrain = data{i,1};
if (nstrain(1) ~= 'S')
continue;
end
% Get the strain and clone IDs from the name
strains(i,1) = str2double(nstrain(2:4));
strains(i,2) = str2double(nstrain(6:8));
% The parent ID
parent = data{i, 3};
strains(i,3) = str2double(parent);
if isnan(strains(i,3))
strains(i,3) = roman2double(parent);
end
% The slide ID
slide = data{i, 2};
strains(i,4) = str2double(slide);
if isnan(strains(i,4))
strains(i,4) = roman2double(slide);
end
% And the record creation date as initial creation date
val = data{i, end};
[curr_time, indx] = strptime(val, '%m/%d/%Y %H;%M');
if (indx > length(val))
strains(i, 5) = mktime(curr_time);
end
% Now we process the comments
% Start by splitting them
comments = strsplit(data{i,4}, ';');
% We match the expression of interest once
deads = regexpi(comments, '(dead)|(empty)|(sacrificed)|(removed)|(fixation)|(slipped)|(missing)|(died)|(death)|(virtual)|(isolated)|(deattached)');
clone = regexpi(comments, '(subcloned)|(sub.cloned)|(positioned)|(repositioned)|(moved)');
% Loop over each comment
for j=1:length(comments)
% Temporary variables in case of no match
indx = NaN;
num = NaN;
is_date = false;
% Loop over each word of the comment
words = strsplit(comments{j});
for k=1:length(words)
% Word cleanup to maximize date matching
w = words{k};
if (~isempty(w) && w(1)=='(')
w = w(2:end-1);
end
if (~isempty(w) && w(end)==')')
w = w(1:end-1);
end
if (~isempty(w) && w(end)=='/')
w = w(1:end-1);
end
% Short values might be number
nchars = length(w);
if (nchars<4)
tmp = str2double(w);
if (~isnan(tmp))
num = tmp;
else
tmp = roman2double(w);
if (~isnan(tmp))
num = tmp;
end
end
% Longer ones could be dates, in several formats
elseif (nchars>7)
[curr_time, indx] = strptime(w, '%d/%m/%y');
if (indx <= nchars)
[curr_time, indx] = strptime(w, '%d/%m/%Y');
if (indx <= nchars)
[curr_time, indx] = strptime(w, '%m/%d/%Y');
if (indx <= nchars)
[curr_time, indx] = strptime(w, '%m/%d/%y');
end
end
end
is_date = (indx > nchars);
end
end
% If we found a mention of colony death
if (length(deads{j}>0))
% Store the death date if any found
tmptime = mktime(curr_time);
if (is_date && ~(strains(i,6) > tmptime))
strains(i,6) = mktime(curr_time);
end
% And the slide number if any
if (~isnan(num) && strains(i,4)==0)
strains(i,4) = -num;
end
% If it mentioned subcloning
elseif length(clone{j}>0)
% Store the date if it is anterior to the creation of the record
tmptime = mktime(curr_time);
if (is_date && strains(i,5) > tmptime)
strains(i,5) = tmptime;
end
end
end
if (strains(i,1)==131 && strains(i,2)==-1)
disp(data(i,:));
if (isnan(strains(i,6)))
disp(['-> ' strftime('%d/%m/%y', localtime(strains(i,5))) ' : NaN']);
else
disp(['-> ' strftime('%d/%m/%y', localtime(strains(i,5))) ' : ' strftime('%d/%m/%y', localtime(strains(i,6))) ]);
end
keyboard
end
end
end
% Clean and sort the output
strains = strains(~isnan(strains(:,1)),:);
[junk, indx] = sortrows(strains(:,1:2));
strains = strains(indx,:);
% Try to infer a minimal death date based on subcloning information
unknown = (strains(:,4)<0 & isnan(strains(:,6)));
ids = abs(strains(unknown, [1 4 5]));
subclones = strains(~isnan(strains(:,3)), [1 3 5]);
% Loop over each missing value and check if some clones were subcloned from it
alive = NaN(size(ids, 1), 1);
for i=1:numel(alive)
hits = (subclones(:,1)==ids(i,1) & subclones(:,2)==ids(i,2));
tmax = max(subclones(hits, 3));
% Make sure we get something valid
if (~isempty(tmax) && ids(i,3) < tmax)
alive(i) = tmax;
end
end
% Copy the results
strains(unknown, 6) = alive;
return;
end
% Special conversion for the large slides with roman numbering.
% Switching back to arabic numbers should not be an issue as strains won't cross.
function val = roman2double(chars)
val = NaN;
if (~isempty(chars))
switch chars
case 'I'
val = 1;
case 'DDI'
val = 11;
case {'II', 'IIa'}
val = 2;
case 'III'
val = 3;
case 'IV'
val = 4;
case 'XV'
val = 15;
end
end
return;
end
c4science · Help