Figure 13.

plot of connectivity scores in increasing order for LE weights, LE cluster 22 We aim to further improve the biological specificity of cluster 22 derived from Laplacian Eigenmaps + k-means. To identify co-varying genes with high connectivity in the regulatory network of cluster 22, we measure connectivity by means of the weight matrix D in (2). The connectivity of Affymetrix probes within LE+KM cluster 22 are shown. The most highly connected genes in cluster 22 are Etv3, Zfp386, Kdm4c, Eea1, Fyttd1, which can be used as labels in Schroedinger Eigenmaps.