For loop using Struct data with multiple values per timepoint

Hello,

I am new to matlab so bear with me:

I currently have a 1x1 struct called "XY11" that contains 3 fields "DNA", "EGFRN" and "EGFRC" which are each 29890x1 structs with multiple fields. Particularly, I want to extract the "Integrated Intensity" and "Timepoint" fields, so I can plot the intensity of my objects over time.

Problem is, there are about 200-300 intensities per timepoint. There are 145 timepoints. I've written code to identify and average the intensities for a given timepoint, but I don't know how to structure the For loop so that each of my 145 timepoints is seperated into their own structure. See the code below.

% Identify all timepoints in the DNA struct

time = [XY11.DNA.Timepoint];

% Identify all intensities in the DNA struct

intensity = [XY11.DNA.IntegratedIntensity];

% Set time == 1 to evaluate intensities found in the first timepoint out of 145

timepoint1 = time == 1;

% Create a structure containing all of these timepoints

% Essentially a Struct containing only the number '1', but it helps me keep my timepoints organized

str.time = time(timepoint1);

% Create a structure containing all of the intensities found at timepoint 1

str.intensity = intensity(timepoint1);

% Find the mean of time (1, or whatever time ==) and intensity

means = structfun(@mean, str, 'uniform', 0);

% Find the STD of time (0, because it is unchaning) and intensity

stds = structfun(@std, str, 'uniform', 0);

The output of this code, Struct "str" is attached.

I don't know how to do the "For j = 1:num" thing, because I don't know how to define num. In my XY11.DNA structure there are hundreds of intensities per timepoint, so I'm not sure how to get matlab to know I need it to loop through all 145 timepoints. Maybe I'm just having "writer's block".

Your assistance is greatly appreciated.

[Jiri Hajek]

Hi, maybe you are not aware of the size function, to give you the "num"?

[Ryan Hecksel]

Hi Jiri, thanks for your response.

size(XY11.DNA.Timepoint0; returns 29890 colums each filled with '1'

size(str.time); returns 1 and 214, the number of intensities found in timepoint 1

time = [XY11.DNA.Timepoint];

intensity = [XY11.DNA.IntegratedIntensity];

for i=1:145

timepoint(i) = time == (i)

str.time = time(timepoint(i))

str.intensity = intensity(timepoint(i))

end

I need i to loop through 145 times, where time == i, but when I run this, I get "Unable to perform assignment because the indices on the left side are not compatible with the size of the right side."

Answer 1

[Stephen23]

This would be a thousand times easier to answer if you had uploaded some sample data in the original format.

Instead of actually helping you, I will have to first try and replicate your data structure. Because your explanation is incomplete, I will assume that all 'IntegratedIntensity' and 'Timepoint' data are scalar numeric, and even then my structure might be wrong: much better would be, if you uploaded a sample data structure (which does not have to be all of it, just enough to work with).

As an aside, that data structure would be much much better as 1x1 structures containing 29890x1 numeric fields, rather than 29890x1 structures containing 1x1 fields.

% Fake data structure:
N = 29890;
S.DNA = cell2struct(num2cell([123*rand(N,1),randi(145,N,1)]),{'IntegratedIntensity','Timepoint'},2);
S.ABC = cell2struct(num2cell([123*rand(N,1),randi(145,N,1)]),{'IntegratedIntensity','Timepoint'},2);
S.XYZ = cell2struct(num2cell([123*rand(N,1),randi(145,N,1)]),{'IntegratedIntensity','Timepoint'},2)

S = struct with fields:
    DNA: [29890×1 struct]
    ABC: [29890×1 struct]
    XYZ: [29890×1 struct]

S.DNA

ans = 29890×1 struct array with fields:
    IntegratedIntensity
    Timepoint

% Get unique times:
iin = [S.DNA.IntegratedIntensity];
tpt = [S.DNA.Timepoint];
[uni,~,idx] = unique(tpt);
% Means and standard deviations:
allmeans = accumarray(idx,iin(:),[],@mean); % or GROUPSUMMARY
allstdev = accumarray(idx,iin(:),[],@std);  % or GROUPSUMMARY
%
plot(uni,allmeans, uni,allstdev)
legend({'Means','Standard Deviations'})

[Ryan Hecksel]

Thank you so much for the help!

Answer 2

[Eric Delgado]

I don't know if I understand your issue, but...

uniqueTimePoints = unique(XY11.DNA.Timepoint);

for ii = 1:numel(uniqueTimePoints)
    idx = XY11.DNA.Timepoint == uniqueTimePoints(ii);
    
    str(ii).time      = uniqueTimePoints(ii);
    str(ii).intensity = intensity(idx);
    str(ii).mean      = mean(str(ii).intensity);
    str(ii).std       = std(str(ii).intensity);    
end

[Ryan Hecksel]

I think that will work, but for some reason, uniqueTimePoints = unique(XY11.DNA.Timepoint) return an error:

Error using unique

Too many input arguments.

[Eric Delgado]

What’s the class of XY11.DNA.Timepoint?

[Ryan Hecksel]

Hi Eric,

Thanks for the help. XY11 is a struct containing 3 structs, DNA, EGFRN, and EGFRC. Each of those is a 29890x1 struct.

[Stephen23]

"but for some reason, uniqueTimePoints = unique(XY11.DNA.Timepoint) return an error: "

It returns an error because Eric Delgado's code provides 29890 inputs to UNIQUE, and is equivalent to doing this:

unique(XY11.DNA(1).Timepoint, XY11.DNA(2).Timepoint, .. , XY11.DNA(29890).Timepoint)

Clearly that is not going to do anything useful for you. As the original question shows, using concatenation is one way to handle that comma-separated list correctly. See also:

https://www.mathworks.com/matlabcentral/answers/1656435-tutorial-comma-separated-lists-and-how-to-use-them

Answer 3

[Seth Furman]

Alternatively, you can convert your struct into a table using struct2table.

https://www.mathworks.com/help/matlab/ref/struct2table.html

load str.mat
t = struct2table(str)

t = 1×2 table
        time         intensity  
    ____________    ____________

    1×214 double    1×214 double

varNames = t.Properties.VariableNames;
t = varfun(@transpose,t);
t.Properties.VariableNames = varNames

t = 214×2 table
    time    intensity 
    ____    __________

     1      3.2815e+06
     1      2.5435e+06
     1      3.6094e+06
     1      2.8818e+06
     1      3.3884e+06
     1      5.8135e+06
     1      4.6849e+06
     1      5.1159e+06
     1      4.0684e+06
     1      3.6248e+06
     1      2.3786e+06
     1      2.7508e+06
     1       2.784e+06
     1      3.5541e+06
     1      2.5174e+06
     1      2.9252e+06

Then, you can use groupsummary to get the mean for each unique time.

https://www.mathworks.com/help/matlab/ref/double.groupsummary.html

tSummary = groupsummary(t,"time",["mean","std"])

tSummary = 1×4 table
    time    GroupCount    mean_intensity    std_intensity
    ____    __________    ______________    _____________

     1         214          6.7187e+06       3.0742e+06  

stackedplot(tSummary,["mean_intensity","std_intensity"],Marker="*",MarkerEdgeColor="red",XVariable="time")

Aside: You can also convert the time variable to the appropriate duration or datetime if desired.

tSummary.time = seconds(tSummary.time)

tSummary = 1×4 table
    time     GroupCount    mean_intensity    std_intensity
    _____    __________    ______________    _____________

    1 sec       214          6.7187e+06       3.0742e+06  

tSummary = table2timetable(tSummary)

tSummary = 1×3 timetable
    time     GroupCount    mean_intensity    std_intensity
    _____    __________    ______________    _____________

    1 sec       214          6.7187e+06       3.0742e+06  

stackedplot(tSummary,["mean_intensity","std_intensity"],Marker="*",MarkerEdgeColor="red")