A simple model of the modular structure of transcriptional regulation in yeast

Resolving the general organizational principles that govern the interactions during transcriptional gene regulation has great relevance for understanding disease progression, biofabrication, and biological systems in general. The available genome-level monitoring technologies and the best understood biological work on gene regulation are together providing us with unprecedented amounts of data and universal modeling frameworks in which to reason about regulatory systems on a computational level. Gene regulatory systems exhibit modularity in their regulatory sequences as well as in the corresponding gene expression. This modularity has a nontrivial, general combinatorial structure that can be studied and generalized to model classes of regulatory systems. Here, we study computationally the combinatorial nature of transcriptional regulation by assuming a one-to-one relationship between shared patterns in genome-wide gene-expression and cis-region modules. In our combinatorial framework, the DNA binding events are complementary to their expression counterparts, and together let us approximate the underlying regulation structure. Our model maps regulatory systems onto hierarchical structures which can be approximated by conflating existing large scale gene expression and ChIP-chip data. We have developed methods for building regulatory hierarchies and identifying the basic functional units, or modules, of transcriptional regulation. We validate our model using yeast data by showing agreement of our predictions with experimental data, and using the hierarchies to resolve a finer structure of co-regulation.