FR-like binding repeats in the human genome

by Mr Aymeric Fouquier d'Herouel (KTH Computational Biology)

Tuesday 6 Apr 2010, 13:30 → 14:30 Europe/Stockholm

122:028

I will talk about the results of a study on the Epstein Barr Virus (EBV), where we applied a nearest neighbor statistical model to identify putative human binding regions for a major viral transcription factor. EBV is widely spread in the human population. EBV nuclear antigen 1 (EBNA1) is a transcription factor that activates viral genes and is necessary for viral replication and partitioning, which binds the EBV genome cooperatively. We identified similar EBNA1 repeat binding sites in the human genome using a nearest-neighbour positional weight matrix. Previously experimentally verified EBNA1 sites in the human genome are successfully recovered by our approach. Most importantly, 40 novel regions are found in the human genome, constituted of tandem-repeated binding sites for EBNA1. Genes located in vicinity of these regions are presented as possible targets for EBNA1-mediated regulation. Among these, four are discussed in more detail: IQCB1, IMPG1, IRF2BP and TPO. Incorporating the cooperative actions of EBNA1 is essential when identifying regulatory regions in the human genome and we believe these findings presented to be highly valuable for the understanding of EBV-induced phenotypic changes. Finally, I will discuss the improvements achieved by including higher order statistics when modeling transcription factor binding sites in practical cases.