Fig. 2

Derivation of score cutoff of clusters. a For an Insilco data with fixed noise level, the size and correlation coefficient of output clusters for each of the 10 runs from the algorithm are overlaid on each other and are shown in first subplot. Three subplots correspond to three noise levels in input data matrices. And the cutoff line is shown in green. The clusters sizes and correlation coefficient plot doesn’t change much for different noise levels. b The value of c* as a function of number of conditions in input matrix is shown showing no significant change with number of conditions as shown by the fit with R2 = 0.3 suggesting a good fit as straight line with slope zero and intercept as mean. c The value of c* as a function of number of genes in input matrix is shown showing no significant change of c* with number of genes as shown by the fit with R2 = 0.0290 suggesting a good fit as straight line with slope zero and intercept as mean