Search | Site Map | Contact

Research: Project

Systems biology of flowering and seed formation

Evangelia Dougali, VIB, University of Ghent

Deciphering transcriptional regulatory networks in the context of flowering and seed formation is a very important step to understand the biological processes behind plant reproduction and seed formation and to improve crop breeding and agricultural productivity. In order to study the systems biology of flowering and seed formation, we will first use methods to automatically extract expression modules and their regulators from gene expression compendia. We will apply two different formalisms, each one focusing on different aspects of gene expression data analysis.

The ENIGMA algorithm (Maere et al. 2008) finds sets of genes (modules) that are significantly co-regulated in expression under specific sets of conditions, using perturbation gene experiment profiles as input. The main strength of the ENIGMA algorithm is that it is also able to uncover crosstalk between cellular processes i.e. genes that show expression behavior similar to one module under certain experimental conditions, and to another under different conditions.

The LeMoNe algorithm (Michoel et al. 2007, Joshi et al. 2009, Michoel et al. 2009) uses Bayesian statistics and Gibbs sampling to infer robust expression modules and their potential regulators from microarray data. LeMoNe has been successfully applied to infer regulatory networks in higher eukaryotic systems, e.g. in C. elegans (fig.1,2) to study metazoan development (Vermeirssen et al. 2009) and in human to study, and experimentally confirm, microRNA regulated modules (Bonnet et al. 2010).

Figure 1: Predicted regulators and module genes for module 142 in C. elegans. This module is involved in energy metabolism, growth and ion transport. Yellow: upregulated genes; blue: downregulated genes [6].

Figure 1


Figure 2: The transcription regulatory network for module 142. The kinase CST-1 is predicted as top regulator. Experimental conditions in the condition clusters, external biological data and a reported regulatory interaction (RI) point to DAF-16 as intermediate transcription regulator. CLR predicted, reported regulatory (RI) and yeast one-hybrid (Y1H) interactions indicate that DAF-3 could also play a role in the path from CST-1 to the module genes [6].

Figure 2

Recent application of ENIGMA and LeMoNe to a compendium of in-house and publicly available microarray data focused on abiotic stress in Arabidopsis yielded several modules with yet uncharacterized genes and novel regulators that could be linked to the process of abiotic stress (unpublished data).

We apply these software tools to “omics” data to identify co-expression modules and regulators involved in flower development and seed formation. Most of these regulators are indirect which means that they act through a hidden path of physical interactions. By integrating different types of interactions, we aim to infer direct and biologically relevant pathways in an automatic way in order to uncover the hidden paths between the regulators and co-expression modules.

In a second step, we will apply classical motif finding algorithms to identify cis-regulatory elements in the co-expression modules, after which they will be evaluated for evolutionary conservation. This not only provides extra evidence of the co-regulation of the genes within a module, but also points to the direct regulator of that module. Finally, all these data will be integrated with other functional genomic data, already available or generated within the course of the SYSFLO project, in order to obtain more biologically relevant information on gene regulation in flower development. Especially, we aim to characterize the regulatory paths between predicted regulators and corresponding co-expression modules by integrating Chip-seq data generated by the other partners in the consortium and already known regulatory interactions.


[1] Bonnet, E., Tatari, M., Joshi, A., Michoel, T., Marchal, K., Berx, G., Van de Peer, Y. (2010) Module network inference from a cancer gene expression data set identifies microRNA regulated modules. PLOS One 5(4), e10162.

[2] Joshi, A., De Smet, R., Marchal, K., Van de Peer, Y., Michoel, T. (2009) Module networks revisited: computational assessment of model predictions. Bioinformatics 25, 490-6.

[3] Maere, S., Van Dijck, P., Kuiper, M. (2008) Extracting expression modules from perturbational gene expression compendia. BMC Systems Biology 2:33.

[4] Michoel, T., Maere, S., Bonnet, E., Joshi, A., Saeys, Y., Van den Bulcke, T., Van Leemput, K., van Remortel, P., Kuiper, M., Marchal, K., Van de Peer, Y. (2007) Validating module network learning algorithms using simulated data. BMC Bioinformatics 8, S5.

[5] Michoel, T., De Smet, R., Joshi, A., Van de Peer, Y., Marchal, K. (2009) Comparative analysis of module-based versus direct methods for reverse-engineering transcriptional regulatory networks. BMC Systems Biology 3, 49.

[6] Vermeirssen, V., Joshi, A., Michoel, T., Bonnet, E., Casneuf, T., Van de Peer, Y. (2009) Transcription regulatory networks in Caenorhabditis elegans inferred through reverse-engineering of gene expression profiles constitute biological hypotheses for metazoan development. Molecular BioSystems 5, 1817-30.