(See supplementary note for a discussion on these assumptions).
The clustering procedure is repeated also for S.cerevisiae, this time merging the two datasets and choosing a lower threshold in order to make an unbiased comparison with previous results for E.coli (similar number of edges, see Supplementary Notes S1 for details).
All other parameters were varied over their full range to produce confidence intervals around our different scenarios (see the Supplementary Notes S2 for detailed information on sensitivity analysis and Table S1 for parameter ranges).
See Supplementary Notes S1 for details.
For E.coli, the clustered expression correlations reproduce faithfully a large part of the collection of PC, and the matching clusters-PC is quasi-monogamous (see Fig. 2 and Supplementary Notes S8 for details and statistics).
The semantic similarity measure of Fig. 6 is drawn from [45], and the associated p.value by means of a bootstrapping method, see again the Supplementary Notes S1 for full detail.
In order to take into account also the homology and the architecture of the genomes, we considered maps of paralog genes (PAR) [15] and, for E.coli alone, a map of transcription units (TU) describing the operonal structure of the prokaryotic DNA (see Tables (a) and (b) of Fig. 1 and Supplementary Notes S1 for details and data sources).
Assuming no prior knowledge, a network structure can be inferred solely from microarray data by means of a genome-wide "similarity matrix" [29] (see Supplementary Notes S1 for definitions and algorithms) and used to test which of the types of interactions listed in Fig. 1 emerge significantly.
A similar (even better) unambiguous correspondence is detected between the clusters and the TU (see Supplementary Notes S9), while for MP the percentages are lower but still significant.
In particular, we would like to thank both anonymous reviewers for the valuable comments and questions, the supplementary notes and the constructive recommendations for revising our publication.
