Skip to main content
Figure 8 | BMC Bioinformatics

Figure 8

From: Post hoc pattern matching: assigning significance to statistically defined expression patterns in single channel microarray data

Figure 8

K-Means Support (KMS). The KMS algorithm with 100 iterations and Pearson correlation for distance, established eight 'consensus clusters' that appeared in at least 80 iterations. For each panel, individual observations are plotted along the X-axis and standardized signal intensity averaged for genes in the cluster, is plotted along the Y-axis. For reference, a '0 line' has been added indicating baseline (scale bar = one standard deviation). Within each panel, the members of the cluster are indicated. Panels A, B, E) These clusters show an apparent discrimination between young subjects and all others. Panel C) The cluster most analogous to a monotonic increase. Panels D & F-H) These clusters appear to be isolating patterns based on non-treatment effects. Inset: Figure of Merit (FOM) procedures were used to estimate number of clusters for K-means but results were difficult to interpret.

Back to article page