Vermeirssen Vanessa, Joshi Anagha, Michoel Tom, Bonnet Eric, Casneuf Tine, Van de Peer Yves
Department of Plant Systems Biology, VIB, B-9052 Ghent, Belgium.
Mol Biosyst. 2009 Dec;5(12):1817-30. doi: 10.1039/B908108a. Epub 2009 Jul 17.
Differential gene expression governs the development, function and pathology of multicellular organisms. Transcription regulatory networks study differential gene expression at a systems level by mapping the interactions between regulatory proteins and target genes. While microarray transcription profiles are the most abundant data for gene expression, it remains challenging to correctly infer the underlying transcription regulatory networks. The reverse-engineering algorithm LeMoNe (learning module networks) uses gene expression profiles to extract ensemble transcription regulatory networks of coexpression modules and their prioritized regulators. Here we apply LeMoNe to a compendium of microarray studies of the worm Caenorhabditis elegans. We obtain 248 modules with a regulation program for 5020 genes and 426 regulators and a total of 24 012 predicted transcription regulatory interactions. Through GO enrichment analysis, comparison with the gene-gene association network WormNet and integration of other biological data, we show that LeMoNe identifies functionally coherent coexpression modules and prioritizes regulators that relate to similar biological processes as the module genes. Furthermore, we can predict new functional relationships for uncharacterized genes and regulators. Based on modules involved in molting, meiosis and oogenesis, ciliated sensory neurons and mitochondrial metabolism, we illustrate the value of LeMoNe as a biological hypothesis generator for differential gene expression in greater detail. In conclusion, through reverse-engineering of C. elegans expression data, we obtained transcription regulatory networks that can provide further insight into metazoan development.
差异基因表达决定了多细胞生物的发育、功能和病理过程。转录调控网络通过绘制调控蛋白与靶基因之间的相互作用，在系统水平上研究差异基因表达。虽然微阵列转录谱是基因表达最丰富的数据，但正确推断潜在的转录调控网络仍然具有挑战性。逆向工程算法LeMoNe（学习模块网络）利用基因表达谱来提取共表达模块及其优先调控因子的整体转录调控网络。在这里，我们将LeMoNe应用于线虫秀丽隐杆线虫的微阵列研究汇编。我们获得了248个模块，其中包含5020个基因和426个调控因子的调控程序，以及总共24012个预测的转录调控相互作用。通过基因本体论（GO）富集分析、与基因-基因关联网络WormNet的比较以及其他生物学数据的整合，我们表明LeMoNe识别出功能上连贯的共表达模块，并优先选择与模块基因相关的、涉及相似生物学过程的调控因子。此外，我们可以预测未表征基因和调控因子的新功能关系。基于参与蜕皮、减数分裂和卵子发生、纤毛感觉神经元和线粒体代谢的模块，我们更详细地说明了LeMoNe作为差异基因表达生物学假设生成器的价值。总之，通过对秀丽隐杆线虫表达数据的逆向工程，我们获得了转录调控网络，这可以为后生动物发育提供进一步的见解。
