Error: Attempted to access data(2,:); index out of bounds because size(data)=[1,7129].

Question

I have a microarray data of 38 row and 7130 columns. i am teying to read the data but keeping having the above error.

I debugged and found when i read the data, i have a 1x7129 instead of a 38x7130. I don't know why. my 7130th column contains letters while the rest of the data are numbers. Any idea why this is happening?

Answer 1

Try this code to read the data:

filename = 'yourfilename.txt';
fid = fopen(filename,'r');

% If you have a line with column headers use those 3 lines. Comment if not.
colnames = fgetl(fid);
colnames = textscan(colnames, '%s','delimiter','\t');
colnames = colnames{:};

% Reading the data
tsformat = [repmat('%f ',1,7129) '%s'];
datafromfile = textscan(fid,tsformat,'delimiter','\t','CollectOutput',1);
fclose(fid);

% Get the data from the cell array    
data = datafromfile{1};
labels = datafromfile{2};

---

EDIT To separate your dataset to training and test, do something like this:

train_samp = 1:19;
test_samp = 20:38;
train_data = data(train_samp,:);
test_data = data(test_samp,:);
train_label = labels(train_samp);
test_label = labels(test_samp);

You can also separate samples randomly:

samp_num = size(data,1);
test_num = 19;
randorder = randperm(samp_num);
train_samp = randorder(test_num+1:samp_num);
test_samp = randorder(1:test_num);

I haven't done transposition data = data';. If you have to, just switch row and column indexes in the above code:

train_data = data(:,train_samp);
test_data = data(:,test_samp);

Answer 2

My file is in text tab delimited and here is my code for reading the file:

clear; 
fn=32; 
col=fn+1; 
cluster=2; 
num_eachClass=3564; 
row=num_eachClass*cluster; 
fid1 = fopen('data.txt', 'r'); 
txt_format=''; 
for t=1:col txt_format=[txt_format '%g ']; 
end 
data = fscanf(fid1,txt_format,[col row]); 
data = data'; fclose(fid1); 

Answer 3

Here is my modified code with Yuk's code but i still have error:

clear;
clc;
fn=7129;
col=fn+1;
cluster=2;
num_eachClass=19;
row=num_eachClass*cluster;
fid = fopen('data.txt', 'r');

% Reading the data 
tsformat = [repmat('%f ',1,7129) '%s']; 
data= textscan(fid,tsformat,'delimiter','\t'); 
data = data';
fclose(fid); 


test_num = 0;
train_num = num_eachClass-test_num;
t1=1;
t2=1;
for i=1:row
   t=mod(i-1,num_eachClass); 
   if t<train_num
      Training(t1,:)=data(i,:);
      t1=t1+1;
   else
      Test(t2,:)=data(i,:);
      t2=t2+1;
   end
end
TrainingLabel = Training(:,col);
Training = Training(:,1:fn);
if(test_num~=0)
    TestLabel = Test(:,col);  
    Test = Test(:,1:fn);
end

Here is the error: ??? Index exceeds matrix dimensions.

Error in ==> test at 38

TrainingLabel = Training(:,col);

Answer 4

Hi Experts, i have changed my ID from Mola to Moladou because my Mola Id is not working Here is my whole code:

clear;
fn=7129;
col=fn+1;
cluster=2;
num_eachClass=19;
row=num_eachClass*cluster;
fid1 = fopen('data', 'r');
txt_format='';
for t=1:col
    txt_format=[txt_format '%g '];
end 
data = fscanf(fid1,txt_format,[col row]);
data = data';
fclose(fid1);

%%-----------  training data and test data ------------
test_num = 0;
train_num = num_eachClass-test_num;
t1=1;
t2=1;
for i=1:row
   t=mod(i-1,num_eachClass); 
   if t<train_num
      Training(t1,:)=data(i,:);
      t1=t1+1;
   else
      Test(t2,:)=data(i,:);
      t2=t2+1;
   end
end
TrainingLabel = Training(:,col);
Training = Training(:,1:fn);
if(test_num~=0)
    TestLabel = Test(:,col);  
    Test = Test(:,1:fn);
end
train_row=t1-1;
test_row=t2-1;
%---------------To scale the data----------------------------------
%---------------normalize  standard deviation to 1 ----------------
% % 
% % for i2=1:fn
% %     mean1=mean(Training(:,i2),1);
% %     std1=std(Training(:,i2),1);
% %     Training(:,i2) = (Training(:,i2)-mean1)/std1;
% %     
% %     if(test_num~=0)
% %        Test(:,i2) = (Test(:,i2)-mean1)/std1;
% %     end
% %     
% % end

%%-------------------initial values-------------------
map_size = 15;
neighborhood = 9 ;
learning_rate = 0.8;
lattice = 'r'; % r : rect , h : hexa
iterations = 100;
%-----------------------------------------------------
%%--- To initiate the map

[Map] = initiate_som_Su(Training,train_row,fn,map_size);

%%--- training process--- 
[Map] = training_Batch(Training,train_row,fn,Map,map_size,neighborhood,lattice,iterations);

%--- calibrate the Map
[Map LabelMap] = calibrateMap(Training,TrainingLabel,train_row,fn,Map,map_size);


%% read the data
sD = som_read_data('data.txt');

% filter columns by variance
%% filter columns by variance
var_values = var(sD.data,1);
var_th = 1e5; % just an example that worked for me
indexToRemove = var_values < var_th;
sD.data(:,indexToRemove) = []; % filter out
sD.comp_names(indexToRemove) = [];
sD.comp_norm(indexToRemove) = [];

%% normalize the data
sD = som_normalize(sD,'var');

%% create, initialize and train a SOM 
sM = som_make(sD,'lettice','hexa');
sM = som_autolabel(sM,sD,'vote'); 

%% U-matrix
U=som_umat(sM,'rect');
% U1=U(1:19,1:19);
figure, som_show(sM,'umat','all') 



%% P-matrix
[pheight rad_real perc] = somvis_p_matrix(sM,sD);
tit = sprintf('P-Matrix (radius = %0.2f)', rad_real);
figure, som_show (sM, 'color', {pheight, tit});

%% U*-matrix
[ustar rad_real perc] = somvis_ustar_matrix(sM, sD);
tit = sprintf('U*-Matrix (radius = %0.2f)', rad_real);
figure, som_show (sM, 'color', {ustar, tit}); 

%---Show the labels of map 
% showMap(Map,LabelMap,map_size,data,col);

%---
LabelMap

you can check to see where the error lies and edit that part for me. Thanks