Figure 2.

Representing transcriptomic profiles and gene signatures. (a) A transcriptomic profile $\mathit{x}$ shown as a vector of log2 transformed fold-change (L2FC) values for each gene (blue and red colors represent down- and up-regulation, respectively). (b) An extreme transcriptomic profile ${\mathit{x}}^n$ is defined by selecting the $n$ most up- and down-regulated genes in $\mathit{x}$ (shown as red and blue squares, respectively). (c) A directional signature ($D{S}^n$) is defined by transforming all up- and down-regulated genes in ${\mathit{x}}^n$ to 1 and −1, respectively. The directional signature ($D{S}^n=\{S^{+},S^{-}\}$), is signified by a set of up-regulated genes ($S^{+}$ shown as red circles) and a set of down-regulated genes ($S^{-}$ shown as blue circles). (d) A non-directional signature ($S^n$) is derived from $D{S}^n$ by ignoring the direction of expression changes for all genes (all genes are shown as black circles). (e) A pathway containing a collection of proteins can be represented by a set of genes (which encode the proteins) and defined as a non-directional signature ($S$). (f) A causal network comprised of interacting proteins can be represented simply by a collection of genes (which can be represented as $S$. (g) A transcriptomic database is a collection of $\mathit{x}$. (h) A bioactivity signature database is a collection formed by one or more of the following types of signatures: ${\mathit{x}}^n,DS,D{S}^n,S$, and $S^n$.