More power via graph-structured tests for differential expression of gene networks

Abstract

We consider multivariate two-sample tests of means, where the location shift between the two populations is expected to be related to a known graph structure. An important application of such tests is the detection of differentially expressed genes between two patient populations, as shifts in expression levels are expected to be coherent with the structure of graphs reflecting gene properties such as biological process, molecular function, regulation or metabolism. For a fixed graph of interest, we demonstrate that accounting for graph structure can yield more powerful tests under the assumption of smooth distribution shift on the graph. We also investigate the identification of nonhomogeneous subgraphs of a given large graph, which poses both computational and multiple hypothesis testing problems. The relevance and benefits of the proposed approach are illustrated on synthetic data and on breast and bladder cancer gene expression data analyzed in the context of KEGG and NCI pathways.

Inferring the functional effect of gene expression changes in signaling pathways

“…However, the practical applicability of such tools is limited given that signaling is a complex interplay between gene activity and pathway topology with final consequences of difficult evaluation. On the other hand, tools that offer a more sophisticated analysis by means of any type of modeling are stand-alone applications written in the statistical language R of difficult use for non-expert users and provide limited graphical representations (11,12). …”

“…However, they are purely descriptive, or, in the best case, they produce a global statistic similar to an enrichment analysis (10). Only a few tools have been recently published addressing the problem of interpreting changes in gene expression within the context of a pathway (11,12). However, these tools have been implemented in the statistical programming language R (http://www.R-project.org), which drastically limits its use only to experienced data analyzers.…”

A transcriptomics model of estrogen action in the ovine fetal hypothalamus: evidence for estrogenic effects of ICI 182,780

“…Then, the DEGraph package (Jacob et al. ) was used to test whether this gene network was differentially expressed between Estradiol, ICI 5 μ g or ICI 5 mg and the control group. As result, significant subnetworks composed by nodes that are colored according to the t ‐statistics (or t ‐score) for the mean difference of expression between the two conditions for each gene were obtained.…”

“…To prove this hypothesis, the KEGG pathway "map04915" (estrogen signaling pathway) was first translated into a gene network with the KEGGraph package (Zhang and Wiemann 2009). Then, the DEGraph package (Jacob et al 2012) was used to test whether this gene network was differentially expressed between Estradiol, ICI 5 lg or ICI 5 mg and the control group. As result, significant subnetworks composed by nodes that are colored according to the t-statistics (or tscore) for the mean difference of expression between the two conditions for each gene were obtained.…”

Analyzing LC/MS Metabolic Profiling Data in the Context of Existing Metabolic Networks

“…A graph-structured two sample test finds if a subgraph is differentially expressed under different treatment conditions 62 . A Markov Random Fields model was proposed to use the pathway structures in identifying differentially expressed genes and important sub-networks 63 .…”