Figure 2

Performance evaluation of the resampling method to identify oscillating transcripts. For LL-, LD- and ODE-based time courses 36 time courses were simulated (original simulations). From these original simulations 3 replicates (initial dataset) for each type of time course were generated by randomly selecting one time point from each of the original simulations to mimic experimental sampling procedures. To generate the averaged dataset, the expression values of the 3 replicates at each time point were averaged. The initial datasets were furthermore used to generate 36 resampled datasets by random sampling at each time point. All datasets were analyzed with ARSER and true positives (A- C) and false positives (D) were calculated requiring increasing consensus between the datasets (see Methods for details). A consensus of 10 thereby means that a gene is found in at least 10 different resampled or originally simulated datasets. For the averaged time course the true and false positives calculated are displayed as a line for comparison.