Skip to main content
Figure 1 | BMC Systems Biology

Figure 1

From: Assessment of network perturbation amplitudes by applying high-throughput data to causal biological networks

Figure 1

Constructing and scoring causal network models. (a) The Selventa Knowledgebase is composed of individual causal relationships that are curated from available sources (generally published scientific literature). For example, in the statement “A → exp(X)” entity A may represent the activity of a particular kinase that has been shown to lead to increased expression of gene X. A HYP is a set of causal relationships derived from the Selventa Knowledgebase that relate an upstream entity to the downstream measurable entities that it regulates. In the HYP example entity A regulates the expressions of genes W, X, Y, and Z following the specific regulation signs “→” (positive regulation) and “--|” (negative regulation). (b) An aggregated HYP can be generated from a causal network model, which is a collection of causally linked entities derived from the Selventa Knowledgebase (grey edges). This causal network model can be augmented with the causal relationships to the downstream measureable entities of its nodes (black edges), also derived from the Selventa Knowledgebase. As a consequence, a causal network model can be equivalently viewed as a collection of causally linked HYPs. Next, all the downstream measurable entities of the network model nodes are combined by adjusting their signs based on the causal relationships in the network. This essential step toward the construction of the aggregated HYP is well-defined as long as the network model is causally consistent (see Methods). Because these nodes have a negative causal relationship with the reference node (node A), the regulation signs “→” or “--|” of the downstream measureable entities from nodes C and D are inverted (red edges) when constructing the aggregated HYP from the causal network model. (c) NPA scores are calculated from high-throughput data, such as differential gene expression data obtained from treatment versus control comparisons, and applied to an NPA scoring algorithm in the context of a specific causal network model represented by its aggregated HYP.

Back to article page