Figure 1.

The key steps in the approach: First, for each candidate hub protein, we identify interaction partners from a human interactome and associate expression levels to each partner. Second, we compute for each protein complex its coordinated expression in all tissues by calculating the average co-expression (PCC) between the hub protein and its interaction partners. Third, we rank for each complex the tissues by their coordinated expression values (PCC scores). According to the underlying hypothesis, the tissue with the highest coordinated expression value will be the one where mutations would most likely cause a disease phenotype.