matlab calls from within fRAT

[schluppeck]

hi elliot,

... looking at the NORDIC code with all the wrinkles, etc. rehashing that in python seems like not the right way to go (as we talked about). To call NORDIC and related matlab functions from within your code would be a nice feature, though ... esp thinking about you working through several datasets using maybe a script to run that.

I had a play around on my macOS laptop to find minmal steps to set up and run and it's not that painful. once you feel a bit better (🦠!) have a look and see what you think...

NIPYPE, matlab setup

minimal working example / matlab

to set up matlab as a command line call, I did:

# ln -s real_path fake_path
sudo ln -s /Applications/MATLAB_R2021a.app/bin/matlab /usr/local/bin

as /usr/local/bin is on my path and the python shell inherits the path variable, this looks like it works. then:

python3

and

# as per doc example with minor mods
import nipype.interfaces.matlab as matlab
mlab = matlab.MatlabCommand(mfile=False)  # don't write script file
# createa  variable x with 100 numbers in it, save out to TXT file
mlab.inputs.script = "x = rand(100,1); save('test.txt','x','-ascii');"
# run the command (this calls w/ -nodesktop and other good options)
out = mlab.run()  

NORDIC test

... working towards running the NORDIC or NORDIC_NIFTI commands you could explore this direction:

# as per doc example with minor mods
import nipype.interfaces.matlab as matlab
mlab = matlab.MatlabCommand(mfile=False)  # don't write script file

# create / path M file for NORDIC run?
# or set a function calle NORDIC_fRAT(infile, outfile) that you can 
# then call?
#
# create a  variable x with 100 numbers in it, save out to TXT file

infile = "'some_nifti_file.nii'" #matlab string inside python string
outfile = "'some_nifti_file_NORDIC.nii'"
params="'a', 1, 'b', 2" # key/value version as per varargin
cmd_string = f"NORDIC_fRAT( {infile}, {outfile}, {params} );"
print(cmd_string)

# set it up for running...
mlab.inputs.script = cmd_string

# run the command (this calls w/ -nodesktop and other good options)
doRun = false  # allow switch on/off

if doRun then:
    out = mlab.run()