Figure 2
From: EDISA: extracting biclusters from multiple time-series of gene expression profiles
Performance of EDISA on an synthetic dataset. EDISA was applied to a synthetic dataset with implanted modules and different levels of noise. The overlap of the implanted modules and the modules mined by EDISA were scored (equation 15). Six runs with 400 iterations were performed, with τ G = 0.1 and τ C = 0.2 for σ ∈ [0,0.5], τ G = 0.15 for σ = 0.7 and τ G = 0.2 for σ = 0.9.