initial measurement of zero to the expression and time for each replicate to capture the initial slope. We then calculated the median slope between each pair of adja cent time points. For a given gene, g, we created a vec tor of median slopes, v, for each profile as the number of time http://www.selleckchem.com/products/Sorafenib-Tosylate.html points, r 1,2.. R, R is the number of replicates, xigr is the expression at time point i for gene g and replicate r and t is the time at time point i. Thus, for n time points, there were n 1 distinct slopes. Maximum and minimum expression ratios The maximum and minimum expression ratios Inhibitors,Modulators,Libraries were important for finding genes with the same magnitude of expression. Biologically, maximum and minimum Inhibitors,Modulators,Libraries expression ratios reflected the impact of signaling via specific transduction pathways and represented the best window of measurement of this change.

These measurements were found from the median ratios over all replicates for a given gene across time points. The maximum expression for a given gene was defined as time points, r 1,2.. R, R is the number of replicates, xigr is the expression at time point i for gene Inhibitors,Modulators,Libraries g and repli cate r. Time to maximum and time to minimum expression Time to minimum and maximum expression and slope between measurements reflect the dynamics of indivi dual gene expression and in many cases where common patterns are observed indicate coordinate control of transcription rates of a group of genes by a common transcription factor. The time of maximum expres sion for a given gene was defined as the i corresponding i 1,2.. n, n is the number of time points, r 1,2..

R, R is the number of replicates, xigr is the expression at time point i for gene g and replicate r. Steepest positive and steepest negative slopes The Inhibitors,Modulators,Libraries steepest positive and negative slopes indicate the maximum rate of over expression and under expression. This feature was selected because it emphasizes these extreme rate changes. The measurements were defined using the median slope as described above and taking the Cilengitide maximum positive slope and the maximum negative slope. Thus, the steepest positive slope for a given gene number of time points, v is the slope between time point i and i 1. Following this, we used the PAM algorithm to cluster the data. Inputs to the algorithm were all of the features described above with equal weight on each. Euclidean distance was used to measure dissimilarity among the selected features.

The number of clusters, k, was determined via the gap statistic. Here, we examined the gap from k 3 15 for both irradiated selleck bio and bystander conditions. The num ber of clusters k is generally chosen where gap gap sk and sk is the estimate of standard deviation from the gap. However, we examined all elbow points on the graphs and presented those that provide the best results in terms of separation of clusters and the homo geneity metric. Evaluating clustering methods In general, cluster validity can be assessed on three bases, within method metrics, between method metrics and clus ter si