Methods blog · Cancer Systems Biology & Drug Resistance
XDaMoSiNType-I and Type-II cross-talks: categorising pathway cross-talk in data-driven signalling networks
A categorisation of signalling cross-talk tailored to data-driven (undirected, context-specific) network models, plus a simple algorithm, XDaMoSiN, to detect both types between any two pathways.
#pathway cross-talk#signalling networks#network topology#systems biology
“Cross-talk” is one of the most used and least precisely defined terms in signalling biology. Existing definitions assume a static, curated, directed pathway map. However, the networks we infer from omics data, such as the p₁-model networks of resistant cancer cells, are context-specific and undirected. In this book chapter (IntechOpen, 2018) we reviewed existing categorisations and proposed one that fits data-driven models.
Two types are enough
Treat a data-driven signalling network as an undirected graph and fix two pathways, \(P_a\) and \(P_b\), with node sets that may overlap. We define:
- Type-I cross-talk. A signalling link \(\{g_i, g_j\}\) connects the two pathways, with \(g_i \in P_a \setminus P_b\) and \(g_j \in P_b \setminus P_a\). Information flows across a bridge edge.
- Type-II cross-talk. Two links \(\{g_i, g_j\}\) and \(\{g_j, g_k\}\) meet at a shared node \(g_j \in P_a \cap P_b\), with \(g_i\) in one pathway and \(g_k\) in the other. The pathways communicate through a common component.
When signalling activities are treated as non-causal relationships, as they are in correlation- or ERGM-based network models, every cross-talk type defined by earlier, topology-based categorisations maps onto Type-I or Type-II. The two-type scheme is therefore complete for this setting while being much simpler to compute.
XDaMoSiN: detecting both types
We also gave a simple, intuitive algorithm, XDaMoSiN (cross-talks in data-driven models of signalling networks). Given a network and two pathways, it:
- partitions nodes into \(P_a\)-only, \(P_b\)-only and shared;
- reports every edge between the two exclusive sets as a Type-I cross-talk; and
- reports every pair of edges meeting at a shared node with endpoints in opposite pathways as a Type-II cross-talk.
Why it is useful
Because the categorisation does not rely on prior directionality, it can be applied to any context-specific network (resistant versus sensitive, tumour versus normal, infected versus uninfected) to reveal latent mechanisms such as acquired drug resistance that emerge from the adaptable, dynamic nature of signal transduction. The same logic underpins our later tools XTalkiiS and rXTalkViz.
Read the paper
-
Cross-talk categorisations in data-driven models of signalling networks: a system-level view
In: Gene Expression and Regulation in Mammalian Cells - Transcription From General Aspects, IntechOpen · 2018