|Institute for Advanced Biosciences||Keio University|
|MathDAMP||Mathematica package for differential analysis of metabolite profiles|
When highlighting differences between metabolite profiles with MathDAMP, arithmetic operations are applied to all corresponding signal intensities from raw datasets on a datapoint-by-datapoint basis.
The datasets have to be normalized in order to allow the direct comparison. For this purpose, a representative set of peaks is picked from every dataset. The peak sets serve the sole purpose of alignment, do not have to contain all the peaks from the datasets and may contain erroneous peak picks as well. Parameters of a time shift function (any mathematical function to model the retention/migration time shifts between samples) are found using a combination of dynamic programming and global optimization. The normalization procedure is able to reliably find the optimal parameters even if the peak sets contain a small number of corresponding peaks.
Various types of differences can be highlighted between the normalized datasets. These include a simple comparison of two datasets, identification of outliers within multiple datasets, comparison of two groups of replicate datasets, and the comparison of multiple groups of replicate datasets. The results are visualized using density plots. Overlaid chromatograms/electropherograms in the vicinities of the most significant differences can be plotted as well.The plots may be annotated using a standard compound library to allow easier identification of the peaks.
For more information regarding the implementation and usage, please refer to the MathDAMP package source and example notebooks in the Examples section.
|Institute for Advanced Biosciences