Static network structure can be used to model the phenotypic effects of perturbations in regulatory networks

Abstract

Motivation: Biological processes are dynamic, whereas the networks that depict them are typically static. Quantitative modeling using differential equations or logic-based functions can offer quantitative predictions of the behavior of biological systems, but they require detailed experimental characterization of interaction kinetics, which is typically unavailable. To determine to what extent complex biological processes can be modeled and analyzed using only the static structure of the network (i.e. the direction and sign of the edges), we attempt to predict the phenotypic effect of perturbations in biological networks from the static network structure. Results: We analyzed three networks from different sources: The EGFR/MAPK and PI3K/AKT network from a detailed experimental study, the TNF regulatory network from the STRING database and a large network of all NCI-curated pathways from the Protein Interaction Database. Altogether, we predicted the effect of 39 perturbations (e.g. by one or two drugs) on 433 target proteins/genes. In up to 82% of the cases, an algorithm that used only the static structure of the network correctly predicted whether any given protein/gene is upregulated or downregulated as a result of perturbations of other proteins/genes. Conclusion: While quantitative modeling requires detailed experimental data and heavy computations, which limit its scalability for large networks, a wiring-based approach can use available data from pathway and interaction databases and may be scalable. These results lay the foundations for a large-scale approach of predicting phenotypes based on the schematic structure of networks.

Boolean ErbB network reconstructions and perturbation simulations reveal individual drug response in different breast cancer cell lines

“…Network models provide novel insight [3,4] and allow us to perform efficiently simulations to predict systems behaviour or evaluate certain hypotheses [5]. Furthermore, combining perturbation experiments with the measurements of system dynamics seems to be even more efficient than time series data on their own [6-8].…”

Quantifying the multi-scale performance of network inference algorithms

“…Our proposals extend scores based on individual edges by considering paths of length > 1. Multi-scale analysis of networks is now an established research field; indeed techniques based on Page et al (1999) have recently come to light in both gene regulatory network (Morrison et al, 2005;Winter et al, 2012) and protein signalling network (Johannes et al, 2010;Feiglin et al, 2012) analyses. This work differs to these within-network studies by focussing on the challenge of performance assessment.…”

“…Our second proposed MSS represents an attempt to explicitly prioritise pairs of vertices which are highly connected over those pairs with are weakly connected: Definition: For each pair (i, j) ∈ V × V we will compute an effect 0 ≤ e ij ≤ 1 that can be thought of as the importance of variable i on the regulation of variable j according to the network G (in a global sense that includes indirect regulation). To achieve this we take inspiration from recent work by Feiglin et al (2012) as well as Morrison et al (2005); Winter et al (2012), who exploit spectral techniques from network theory. Since the effect e ij , which is defined below, includes contributions from all possible paths from i to j in the networkĜ, it explicitly captures differential connectivity.…”

Detecting and Removing Inconsistencies between Experimental Data and Signaling Network Topologies Using Integer Linear Programming on Interaction Graphs

“…Significant work has been published on this front attempting to identify inconsistencies between measured data and signaling topologies [6]–[16]. Some methods also facilitate an optimization of the network structure to identify the wiring diagram that can best fit the data at hand [6], [7], [15].…”