2.4. Finding Master Regulators in Networks
A second workflow was designed to find master regulatory molecules in signal transduction pathways upstream of identified transcription factors. The workflow firstly maps transcription factors to the TRANSPATH® network (BIOBASE) [13] where they are subjected to a master regulator search with a maximum radius of 10 steps upstream of the factor nodes. A new score is assigned to each potential master regulator that reflects its specificity for the downstream effector TFs Equation (7).
(7)   In Equation (7), k is the radius of pathway steps that effector nodes can be separated from the master regulator, Mk is the number of input molecules reached by the regulator within k steps, and Nk is the total number of molecules reached from the master regulator within k steps. The quantities Mmax,k and Nmax,k are the highest values among all possible master regulator nodes and normalize the score to the (0,1)-interval. The higher this score, the more specific this master regulator is for the set of input molecules. The parameter κ is a user-defined penalty, the default of which is set to 0.1.
To make master regulator scores comparable, we compute a Z-score using 1000 randomly sampled molecule sets of the same size as the input set. These are subjected to the search keeping all other conditions as for the original input. By default, the workflow filters master regulator molecules with Z-score of >1.0 and a score of >0.2. Additional steps are performed by the workflow, such as mapping TRANSPATH® entities to both Ensembl Gene IDs and to UniProt protein IDs. The table with Ensembl Gene IDs is further annotated with additional information, gene description and gene symbols. Finally, the table with master regulatory molecules is sorted by the sum of the ranking of both scores, and networks for each master regulator can be visualized as diagrams in the hierarchical layout.