MetaboLab GUI

Metabolab GUI is a graphical user interface for data preprocessing for statistical data analysis (e.g. PLS toolbox). It provides a graphical user interface for spectral alignment, baseline correction, spectral scaling, definition of exclude regions, non-linear scaling and bucketing. If PLS-toolbox is installed the final spectra can be converted into a PLS-toolbox data set which can be read into PLS toolbox. If classes are assigned to spectra, they are translated into the PLS-toolbox data set.

Screen Shot 2012-02-06 at 14.58.07

Classes are assigned by selecting Select spectra/Select from list. Classes are assigned to specific spectra by selecting the spectra and setting the appropriate class.

Screen Shot 2012-02-06 at 14.58.38

Spectra belonging to different classes are plotted in different colours (blue for class=1, red for class=2, black for class=3).

Screen Shot 2012-02-06 at 14.59.49

Baseline correction: For best results, use spline baseline correction with the linear spline option turned on. This option will interpolate to selected baseline points next to each other if they are more than the chosen number of data points apart. This improves baseline correction for parts of the spectrum where there is only very little baseline information available:

Screen Shot 2012-02-06 at 15.01.17


Spectra before (left) and after (right) spline baseline correction.

Metabolab baseline a


Graphic selection of exclusion regions (left click to define exclusion regions, middle click to remove them).

metabolab exclude2


Regions can be sequentially aligned. NMRLab provides an interface to the icoshift software (http://www.models.life.ku.dk/icoshift), which has to be obtained separatly.

segAlign

 

Noise can be removed from the spectrum using the noise filtering option. A noise region and a threshold below everything is discarded can be defined.

Screen Shot 2012-02-06 at 15.07.26


Buckets can be defined either in ppm (left) or in points.

Screen Shot 2012-02-06 at 15.06.03


Advance scaling: Spectra before (left) and after (right) extensive glog scaling:

glog

Using a series of technical replicate samples, the lambda parameter of the glog transform can be optimized automatically. If no optimization is performed a default of 1e-8 is used for lambda. No optimization is performed on y0 (a default of 0 is used).

If the Create PLS dataset option is used, the final data is contained in a global variable called plsData. The specified parameters (including spectra etc.) are contained within the global variable metaboSpc.



Back

© Christian Ludwig 2011