Evaluating network inference methods in terms of their ability to preserve the topology and complexity of genetic networks

Abstract

Network inference is a rapidly advancing field, with new methods being proposed on a regular basis. Understanding the advantages and limitations of different network inference methods is key to their effective application in different circumstances. The common structural properties shared by diverse networks naturally pose a challenge when it comes to devising accurate inference methods, but surprisingly, there is a paucity of comparison and evaluation methods. Historically, every new methodology has only been tested against gold standard (true values) purpose-designed synthetic and real-world (validated) biological networks. In this paper we aim to assess the impact of taking into consideration aspects of topological and information content in the evaluation of the final accuracy of an inference procedure. Specifically, we will compare the best inference methods, in both graph-theoretic and information-theoretic terms, for preserving topological properties and the original information content of synthetic and biological networks. New methods for performance comparison are introduced by borrowing ideas from gene set enrichment analysis and by applying concepts from algorithmic complexity. Experimental results show that no individual algorithm outperforms all others in all cases, and that the challenging and non-trivial nature of network inference is evident in the struggle of some of the algorithms to turn in a performance that is superior to random guesswork. Therefore special care should be taken to suit the method to the purpose at hand. Finally, we show that evaluations from data generated using different underlying topologies have different signatures that can be used to better choose a network reconstruction method.

Immunometabolic Network Interactions of the Kynurenine Pathway in Cutaneous Malignant Melanoma

“…In the correlation network analysis, a node represents a KPMs or immune-related biomarker, and an edge is defined by statistically significant correlations between biomarkers in analyses of pretreatment (PRE) samples. These values (cut-off of 0.5; is the highest level of correlation for which the network is not fragmented) were used to reconstruct networks in PRE and TRM CMM groups (before and during treatment) (49). We defined hubs as nodes with a high degree of centrality (high connectivity with other nodes) in PRE and TRM network.…”

Untitled

“…In recent years, the K(G) and Aut(G) analyses have been applied to a number of biological networks [64, [69][70][71][81][82][83][84][85][86][87][88]. One advantage of Kolmogorov complexity over a purely symmetry description using Aut(G) is that it better captures all the graph structure by measuring all of the non-randomness.…”

High-Dimensional Bayesian Network Inference From Systems Genetics Data Using Genetic Node Ordering

“…Gene network inference is a deeply studied problem in computational biology (Friedman, 2004;Albert, 2007;Bansal et al, 2007;Penfold and Wild, 2011;Emmert-Streib et al, 2012;Marbach et al, 2012;Äijö and Bonneau, 2016;Kiani et al, 2016). Among the many successful methods that have been devised, Bayesian networks are a powerful approach for modelling causal relationships and incorporating prior knowledge (Friedman et al, 2000;Friedman, 2004;Werhli and Husmeier, 2007;Mukherjee and Speed, 2008;Koller and Friedman, 2009;Pearl, 2009).…”