Fig. 1From: Markov chain Monte Carlo for active module identification problemBehavior of subgraph log-likelihood values and ranking AUCs depending on the number of MCMC iterations. A real protein-protein interaction graph of 2,034 vertices is used as G, a module of 200 vertices is chosen uniformly at random. Green line: log-likelihood values for subgraphs Si generated during one MCMC run. Red line: AUC values for rankings based on 1,000 independent MCMC samples depending on the chosen mixing time estimate. Blue line: AUC values for rankings based on one MCMC run calculated on all samples Si for i>25,000Back to article page