Figure 5.
Sensitivity of disease module detection to the interactome. For each interactome, STRING (A) or iRefIndex (B), we first applied each disease module into the original interactome. Then, a fraction of protein–protein interactions (PPIs) randomly selected from the original interactome were removed. Finally, we performed the disease module detection algorithms again using perturbed interactome and compared the fraction of overlapping genes in the union genes of the disease module identified from the original and perturbed interactome, respectively.
