Getting started with the Globe_Toolbox
The Globe_Toolbox provides an interactive environment to globally fit
3-Dimensional data to any desired model. Although it is specifically designed
for the analysis of time-dependent spectroscopic data (i.e. absorption as function
of wavelength and time), it can be used to view and analyse different sorts
of correlated time-based data, such as pH or concentration as function of
time and sample number.
Before you start
Make sure all required toolboxes are installed (See
System Requirements ).
Copy all Globe_Toolbox files to a directory, and properly index this
location and its subfolders for Matlab with File>Set Path. Add the
subfolders you just copied Globe_Toolbox to and save. Type
Globe_Toolbox in the Command Window, and press enter to execute.
Check if the provided Help files are correctly installed by pressing the
Help button in the started GUI. If the documentation is not found,
press Matlab's Start button>Desktop Tools>View Start button configuration files>Refresh
start button. Check if the Globe_Toolbox is listed (referring to the
correct location of the info.xml file). If the toolbox is not listed, try
restarting Matlab, and check again if it is listed. Press Close.
Screen size
The main graphical user interface is started by executing
Globe_toolbox_v1_00 in Matlab's command window. Because screen sizes
and resolutions differ on each machine, it may be possible that the default
settings are not appropriate for your display. The size of the main window
needs to be modified if words appear clipped. If horizontal stretching of
the main interface renders all text properly visible, it may be necessary
to adjust the main window size. Type in the command window:
open Globe_toolbox_v1_00, and press Enter. The Editor should now
open. If not, set your paths first (see above). Adjust the size to match
your display settings by changing the Position paramaters on line 38
(the 4 numbers are referenced to the left bottom corner of screen and in
normalised units -full width of your screen corresponds to 1- and
represent [horizontal_start vertical_start width height]). Change for
instance the width to 0.85, and rerun Globe_toolbox_v1_00 to see if
everything is now visible.
Working Method
Matlab automatically generates the differential equations for the specified
model, making use of its SimBiology Toolbox. The differential equations are
then solved numerically. Consequently, the concentration profiles for each
species or compartment are generated, which in turn are fitted to the data
via a non-linear least-squares algorithm. In other words, the concentration
profiles (which change with rate constants) are iteratively fitted to the
data. The amplitude of the exponential function (characterised by a rate
constant) gives the spectrum (if spectroscopic data are fitted). The
amplitudes are fitted linearly via mldivide.
Non-linear fitting
A least squares fitting method is used to minimise the difference between the data
and the simulated dataset by changing the starting values of the user-given
time constants. In addition, time zero and an offset parameter are optimised
as well. All parameters are passed to Matlab's lsqnonlin optimisation
algorithm (type 'doc lsqnonlin' for more information). Alternatively, a
direct search method can be used to scan the parameter space for a global
minimum. This takes significantly more time, but is less likely to end up
in a local minimum. The patternsearch algorithm is used to achieve
this (type 'doc patternsearch' for more information).
Linear fitting
The spectra are fitting using Matlab's QR decomposition (type 'doc qr' in
command window), which is used to solve linear systems with more equations
than unknowns:
'The qr function performs the orthogonal-triangular decomposition of a matrix.
This factorization is useful for both square and rectangular matrices. It
expresses the matrix as the product of a real complex unitary matrix and an
upper triangular matrix.'
In practice, this means that from input matrix A it produces an upper
triangular matrix R of the same dimension as A and a unitary matrix Q so
that A = Q*R.
Further links