Univariate analysis methods (Fold change, t-test and Volcano plot are suitable only for analysing two groups (see Diet effect or Platform differences in Analysis).
Other methods (Boxplot, Heatmap, PCA, PLS-DA, Chord diagram) accept various combinations of matrices and diets inputs (see Combinations in Analysis).
Example plots are based on real data and support interactive tools for controlling the plot (zoom, pan, hover).
Volcano plot shows significance (t-test p-value) vs. fold change (ratio of group averages). Groups can be defined by different diets for one matrix or different matrices for one diet. Datapoints with high fold change and low p-value represent analytes that have large magnitude changes between groups and are also statistically significant. All p-values are FDR adjusted.
The aim of Partial Least Squares Discriminant Analysis is to achieve maximum separation among classes by rotating the coordinates. It allows visualization of multidimensional dataset and identification of analytes that contribute to the diversity of groups.
Bubble plot is an extended scatter plot. Additional information is represented by the size of the dots. In MetaboAtlas21 bubble plot is used to compare number of carbon atoms and double bonds of analytes in a lipid class. Size of the points represents normalized intensity.