Supplementary Information: LICORN: learning co-operative regulation networks from gene expression data

We used the normalised data set available at http://genome-www.stanford.edu/cellcycle/data/ rawdata/combined.txt. This expression data set [9] comes from yeast cultures synchronised by four independent methods: alpha for alpha factor arrest, elu for elutriation, cdc15 for arrest of a cdc15 temperature-sensitive mutant, cdc28 for arrest of a cdc28 temperature-sensitive mutant (from [2]). As explained in [9], this data set represents log2 ratios that have been normalised so that for each synchronisation method, the average log2 ratio over the course of the experiments is equal to 0. We worked on data for each synchronisation method separately. We made use of the temporal structure of the expression profiles, by calculating the differences between consecutive time points1, and chose a discretisation threshold corresponding to half the overall standard deviation of these differences for the synchronisation method chosen. Note that for a given synchronisation method, the threshold is the same for all genes: