Figure 4.

Using the MADMuscle meta-analysis tool to refine gene expression patterns in a given pathology with a high level of confirmation. A- Heat map showing the differentially expressed genes in DMD muscle from our data set. B- Heat map (merged overlapping clusters) of gene expression in normal and DMD muscle from five independent studies of the database. Samples (columns) were classified according to the experimental group (DMD and normal) and the data set of origin (see color legend). For each gene (rows), the yellow and purple bars indicate the total number of measurements (0 to 5) across studies and the computed p-values from the student's t-test. C- The comparison (A and B) identified 8 different classes of gene expression depending of their status in our data set ("+": over-expression; "-": under-expression) and across the other studies ("+/-": over/under-expression; "NS": no significant differential expression; "NP": no present in the clusters). Among them, the two classes +/+ and -/-, corresponding to genes commonly up- or down-regulated across all the DMD studies, are illustrated by some particular gene names (right hand side). D- Multidimensional scaling (MDS) [83] analysis of gene proximity (genes +/+ and -/-) given by the 1,247 "good" labeled clusters of the database. Results of the projection are represented by a 2 D plot in which green and red indicate gene commonly under and over-expressed in DMD muscle (classes -/- and +/+). Symbols (circle, square or triangle) illustrate particular Gene Ontology (GO) terms associated to these genes (see legend).

Baron et al. BMC Genomics 2011 12:113   doi:10.1186/1471-2164-12-113
Download authors' original image