Fig. 3

Workflow of computational validation. As many as the number of interactions for each time point identified by the proposed method, interactions are randomly selected without considering their time points to construct random networks. From these random networks, a common network is identified. At two time points which can explain phenotypic change, the average difference of perturbation scores is calculated on common network. This random sampling process is repeated as many as 10,000 iterations. Based on the distribution from 10,000 average differences of perturbation scores, the statistical significance of our average difference of perturbation score is performed using statistical test