User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse





Computational Genomic Analysis of Transcriptional Regulation


Lee,  Ho-Joon
Max Planck Society;

External Ressource
No external resources are shared
Fulltext (public)

(Any fulltext), 2MB

Supplementary Material (public)
There is no public supplementary material available

Lee, H.-J. (2008). Computational Genomic Analysis of Transcriptional Regulation. PhD Thesis, Freie Universität Berlin, Berlin.

Cite as: http://hdl.handle.net/11858/00-001M-0000-0010-8004-3
Modern technological advances have been producing a huge amount of highthroughput genome-/proteome-wide data which are to be analyzed for inferring biological knowledge. Computational and statistical analyses are an appropriate and efficient way for such large-scale data analysis. In this thesis we investigate genome-wide transcriptional systems by data integration, which is also a prerequisite for systems biology. Computational and statistical methodologies are developed and applied to heterogeneous genome-wide data sources in a model organism, emph{Saccharomyces cerevisiae}. We aim to discover strong functional signals and related mechanisms from noise-prone genome-scale transcriptional data. First, our analysis starts with groups of genes bound by common transcription factors, called transcriptional modules. They are derived from protein-DNA interaction data and coupled to gene expression and functional annotation data in order to identify functional signals. Standard methods applied to various large-scale gene expression data show that those identified functional modules can be condition-invariant or condition-specific. Second, we extend our module analysis to prioritization of gene regulatory interactions in functional modules identified on a large scale. Our simple integrative approach to such prioritization yields a statistically significant increase of prediction accuracy for two types of reference datasets compared with an original analysis of genome-wide protein-DNA interactions data alone. In addition, our predictions include those regulatory interactions that were not predicted by other algorithms with as good prediction accuracy. Finally, in view of ubiquitous combinatorial regulation by multiple transcription factors, we turn our attention to different sets of target genes in different conditions regulated by pairs of regulators. We develop a method to identify condition-specific co-factors of those regulators that significantly change their target genes in different conditions. We apply the method to genome-wide protein-DNA interactions data generated in diverse cellular conditions. Our predictions include novel cooperative regulator pairs as well as known ones with evidences from gene expression, protein-protein interactions, and conserved motifs data. Further analysis shows that such condition-specific combinatorial regulation occurs more abundantly than expected by chance. In conclusion, our analyses successfully reveal meaningful biological findings and generate concrete hypotheses from heterogeneous genome-wide yeast data. Therefore, this work is expected to contribute as a first step to guiding experimentalists and studying more detailed biological mechanisms.