Abstract

We are developing computational models of transcription and its regulation from high-throughput time course data. I will discuss two models: (1) We model RNA polymerase (pol-II) dynamics using data from pol-II ChIP-Seq time course datasets (wa Maina et al. 2014). Our model is used to estimate the elongation speed and promoter activation kinetics for early targets of ER-alpha in MCF7 cells. By clustering the promoter activation profiles we can associate differences in the kinetics with transcription factor binding, e.g. the earliest clusters are all associated with nearby FOXA1 binding. (2) We model the relationship between pol-II dynamics and mRNA production by combining 3’ pol-II ChIP-Seq data with RNA -Seq data. We find that 11% of genes where we can fit our model exhibit surprisingly long production delays which appear to be associated with splicing. Although transcription delays are longer than splicing delays for most genes, there are a significant set where splicing is the rate-limiting step. We use a Bayesian non-parametric approach to model temporal profiles such as the pol-II activation which allows us to limit the number of free parameters in our model and to capture the uncertainty in our inferences.

C. wa Maina, A. Honkela, F. Matarese, K. Grote, H. Stunnenberg, G. Reid, N. Lawrence and M. Rattray. “Inference of RNA Polymerase II Transcription Dynamics from Chromatin Immunoprecipitation Time Course Data.” PLoS computational biology 10, no. 5 (2014): e1003598.