Fig. 5

Correlation networks between genes (boxes) and metabolomic features from FT-ICR-MS data (ellipses). Positive correlations >0.75 (solid lines) form two groups of sucrose inversion-related entities (marked in yellow and gray, respectively); members from different groups are negatively correlated (dashed lines; a). The global network of topmost positive correlations (>0.85) between genes and metabolomic masses from the non-targeted metabolomics measurements consists of five communities (b). Red and blue colors indicate up- and down-regulation under drought stress, primarily occurring in the communities C3 and C5, respectively. The marked metabolomic mass, a putative glycerophosphoinositol, is the most connected mass within community C4, which in contrast to other communities mainly represents heat-related response (c). The mass and its top correlated gene together separate different environmental conditions, with gradually more induction from control to drought, heat and combined stress (d). A subnetwork of closely interconnected genes and masses (black edges; the ellipses mark fully connected groups) suggests drought-related functions of unassigned masses and unknown genes (e). Known genes related to response to water deprivation are annotated by yellow color, known genes having functions as transport and membrane proteins are annotated by blue color. Gray edges indicate correlations >0.85 to metabolites from GC-MS data; only metabolites with at least ten gene connections are shown. The heatmap shows that stress response profiles are quite homogeneous for all subnetwork members