Fig. 1

Experimentation protocol schematic representation. A collection of microarray dataset is assumed to be generated from multiple, independent studies all following the same experimental protocol and measuring the same quantities. The studies investigate the same biological system regulated by an unknown gene interaction network. The data collection is analyzed with two different approaches, namely Meta-Analysis and Data-Merging. In the first approach correlations among transcription factors and genes are first calculated on each dataset and then summarized, while in the latter the data are merge together, corrected for eventual batch-effects and the correlations are estimated on the pooled data. The correlations retrieved by the two approaches are then compared with a set of known interactions that partly and possibly noisy reconstruct the original gene interaction network. Meta-Analysis and Data-Merging approaches are then evaluated on the basis of their ability of assign highly-significant correlations to known interactions