How can I import text and data from a .txt file?

The format of the file is two column, with headers "Masses" and "Intensities"

However, those headers are rewritten after each scan, so the file looks like this:

Masses   Intensities

10.0     1.2131233e7

11.0     3.13214e5

12.0     2.234234e7

Masses   Intensities

10.0     2.2314324e7

11.0     4.12123123e5

12.0     1.921393e7

Masses   Intensities
...

and so on.

When I use a command like

    A = importdata('C:\Users\gps9h\Dropbox\MBMS\data\2015_8_24\2015_8_24_Run_2_chro.txt','\t'); % code

I get a structure A which has all of the data from the first scan, and also has the headers, which is good. But how do I get it to extract the data from all of the scans? (ie not to stop after it comes across the words 'Masses' and 'Intensities' for the second time?

I played around with fscanf and load commands and nothing was doing quite what I wanted. The IDEAL would be to make a 3-d array where I have scans 1-N, and then a column of masses and a column of intensities for each scan.

Thanks so much! This is my first time using matlab forum so let me know if anything is out of line in how I ask the question or if this has been answered elsewhere.

Answer 1

[Walter Roberson]

vals_per_scan = 3;

fid = fopen('C:\Users\gps9h\Dropbox\MBMS\data\2015_8_24\2015_8_24_Run_2_chro.txt', 'rt');
datacell = textscan(fid, '%f%f', 'CollectOutput', 1, 'Delimiter', '\t', 'CommentStyle', 'Masses');
fclose(fid);

data = permute( reshape(datacell{1}, vals_per_scan, [], 2), [1 3 2]);

This will be a 3D array, vals_per_scan (3) by 2 by the number of scans. If you want the scan number to be the first index, use [2 1 3] instead of [1 3 2]. When deciding which index you want first, please take into account that consecutive locations correspond to changing the first index, so A(2,1) is the item next in memory after A(1,1). If you will be processing scan by scan then it is more efficient to have the scan number last, but if you are processing across scans (e.g., plotting how the first mass changes over the scans) then you should have the fastest changing index first.

This depends upon the number of values per scan always being the same.

[Gregory Simms]

Thank you so much!