Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2011:314-25.
doi: 10.1142/9789814335058_0033.

Defining the players in higher-order networks: predictive modeling for reverse engineering functional influence networks

Jason E McDermott  1 , Michelle Archuleta, Susan L Stevens, Mary P Stenzel-Poore, Antonio Sanfilippo
Affiliations

Defining the players in higher-order networks: predictive modeling for reverse engineering functional influence networks

Jason E McDermott et al. Pac Symp Biocomput. 2011.

Abstract

Determining biological network dependencies that can help predict the behavior of a system given prior observations from high-throughput data is a very valuable but difficult task, especially in the light of the ever-increasing volume of experimental data. Such an endeavor can be greatly enhanced by considering regulatory influences on co-expressed groups of genes representing functional modules, thus constraining the number of parameters in the system. This allows development of network models that are predictive of system dynamics. We first develop a predictive network model of the transcriptomics of whole blood from a mouse model of neuroprotection in ischemic stroke, and show that it can accurately predict system behavior under novel conditions. We then use a network topology approach to expand the set of regulators considered and show that addition of topological bottlenecks improves the performance of the predictive model. Finally, we explore how improvements in definition of functional modules may be achieved through an integration of inferred network relationships and functional relationships defined using Gene Ontology similarity. We show that appropriate integration of these two types of relationships can result in models with improved performance.

PubMed Disclaimer

Figures

Figure 1
Figure 1. Overview of iterative cross-validation predictive modeling approach
1) Network inference from transcriptomic data using CLR. 2) Definition of target clusters for modeling using several partitioning methods. 3) Definition of potential regulators from existing knowledge or topological analysis. 4) Cross-validation of predictive model: A. Divide expression data into related independent groups of observations (i.e. different treatments); B. Build a predictive model using all but one group with the Inferelator; C. Evaluate the performance of the model using the left out group; D. Repeat with next independent group. 5) Use the overall predictive performance to evaluate and refine methods used to determine the network components (targets [2] and regulators [3]).
Figure 2
Figure 2. Performance of a predictive model of neuroprotection and injury during stroke in a mouse model system
A. Target cluster performance. The coexpressed clusters used as targets for modeling are shown (X axis) with bar height (Y axis) indicating the performance (correlation of predicted versus observed expression) for that target in the cross-validation approach. # indicates the poorly performing cluster used in further partitioning and * indicates the accurately predicted cluster shown in panel B. B. Expression of an accurately predicted target. The observed (red line) versus predicted (green line) expression levels (Y axis) for one cluster representing 180 genes is shown over the treatments/time points (X axis). The independent groups used in the cross-validation are indicated in colored boxes, and time points post-treatment (white boxes) and post-stroke induction (grey boxes) are also shown.
Figure 3
Figure 3. Bottlenecks are complementary to transcription factors as candidate regulatory influences
Predictive models were constructed using annotated transcription factors (TFs), topological bottlenecks, or a combination of the two groups (X axis). The mean and standard deviation (error bar) of ten randomly selected sets of genes is shown as a control. Performance (Y axis) using our cross-validation approach indicates that bottlenecks are robustly predictive of system behavior.
Figure 4
Figure 4. Performance of CLR and XOA defined subclusters for prediction
The parent cluster was subclustered using either the CLR (red)- or XOA (blue)- derived associations between genes into the indicated number of subclusters. Performance (mean correlation of observed versus predicted expression levels) is shown on the Y axis. These results support our previous observations that both methods can improve performance.
Figure 5
Figure 5. Comparison of subclustering methods
The mean performance of the methods examined (X axis) across different subclustering levels (3–7 clusters, as in Figure 1) is shown (Y axis). The error bars represent one standard deviation. The methods used are XOA and CLR alone, minimum p value (MinP), maximum p value (MaxP), mean of p values (MeanP) and product of p values (PxP). These results show that combining the CLR and XOA associations using probabilities can improve performance over the individual methods alone, but that only when non-standard methods (maximum p value or mean of p values) are employed to do so.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Donnan GA, et al. Stroke. Lancet. 2008;371(9624):1612–23. - PubMed
    1. Stenzel-Poore MP, et al. Effect of ischaemic preconditioning on genomic response to cerebral ischaemia: similarity to neuroprotective strategies in hibernation and hypoxia-tolerant states. Lancet. 2003;362(9389):1028–37. - PubMed
    1. Marsh B, et al. Systemic lipopolysaccharide protects the brain from ischemic injury by reprogramming the response of the brain to stroke: a critical role for IRF3. J Neurosci. 2009;29(31):9839–49. - PMC - PubMed
    1. Thakar J, et al. Modeling systems-level regulation of host immune responses. PLoS Comput Biol. 2007;3(6):e109. - PMC - PubMed
    1. Oberhardt MA, Palsson BO, Papin JA. Applications of genome-scale metabolic reconstructions. Mol Syst Biol. 2009;5:320. - PMC - PubMed

Publication types

MeSH terms

-