Biology Reference
In-Depth Information
Experiments
FIGURE 6.5
Example of heatmap representation of an expression data set. Here, the heatmap was
generated using the data of
Nicolas et al. (2012)
by reordering both the genes (5875 rows
including newly defined transcribed segments) and the experiments (269 columns)
according to the results of hierarchical clustering (“average link” on the pairwise Euclidean
distance matrix). High expression is represented by darker colours.
The main limitation of the clustering approaches is that they lead to the definition
of clusters that do not overlap, whereas the regulons of the different transcription
factors or the response to various environmental stimuli do overlap. Several
approaches have been proposed to alleviate this constraint. Probably, the most ambi-
tious is the idea of biclustering, whereby attempts are made to find subsets of genes
and conditions that show particularly coherent patterns (
Madeira and Oliveira, 2004;
