Jan Van de Velde1,2,3, Inge De Clercq1,2, Robin Pottie1,2, Xiaopeng Liu1,2, Dries Vaneechoutte1,2,3, Shubhada R. Kulkarni1,2,3, Frank Van Breusegem1,2, Klaas Vandepoele1,2,3 (1) Ghent University, Department of Plant Biotechnology and Bioinformatics, (Technologiepark 927,) 9052 Ghent, Belgium ; (2) VIB Center for Plant Systems Biology, (Technologiepark 927,) 9052 Ghent, Belgium ; (3) Bioinformatics Institute Ghent, Ghent University, Technologiepark 927, 9052 Ghent, Belgium
Abstract of the talk :
Gene regulation is a dynamic process in which transcription factors (TFs) play an important role. While TF binding events can have a direct or indirect effect on the activation or repression of gene transcription, more complex regulation of gene expression is achieved through cooperative binding of different TFs adding an extra combinatorial level of complexity. Despite the functional importance of transcriptional regulation and the fact that >1700 TFs have been identified in the model species Arabidopsis thaliana, our global knowledge about the genes controlled by different TFs is limited. Recently, we have started to unravel the regulatory lexicon of Arabidopsis through the systematic processing of regulatory datasets comprising ChIP-Seq, protein-binding microarrays, chromatin modification data, RNA-Seq data and the delineation of conserved TF binding sites. Through the integration of different experimental and computational input networks using a supervised learning approach we constructed a gene regulatory network (GRN) covering 1,491 TFs and 31,393 target genes (1.7M interactions). We show that this integrated GRN (called iGRN) outperforms all input networks in terms of sensitivity and specificity to predict known regulatory interactions. iGRN has similar performance to recover known functional regulatory interactions compared to experimental methods like Y1H and ChIP. Through the analysis of the predicted target genes for different TFs, we are able to recover for >600 TFs known functions associated to specific biological processes such as growth, development, (a)biotic stress response, hormone signaling, circadian rhythm, etc. Furthermore, the iGRN allowed predicting new gene functions for >1000 TFs. I will present experimental validations for some new TFs predicted to be involved in oxidative stress signaling. In conclusion, the presented iGRN offers a high-quality starting point that integrates different functional omics data sets at the network level to enhance our understanding of gene regulation in plants.