The KM-Algorithm Identifies Regulated Genes in Time Series Expression Data

Figure 4

For nine simulated data sets of different size 𝑛 and length 𝑇, model selection is performed with different maximum biological time lags 𝑝=1,2,3. The results of the model selection process are noted on the 𝑥-axis of each data set (left bar represents true model dimension 𝑚=10, right three bars represent 𝑝=1,2,3 in that order). At the true model dimension 𝑚=10 and the selected dimensions the KM-algorithm is applied five times and regulation results averaged. The “goodness of rank" (GR) measure of the resulting gene ranking is shown.