Integrating mRNA and Polymerase Time Course Data to Model the Dynamics of Transcription

Magnus Rattray

Faculty of Life Sciences, The University of Manchester, Oxford St., Manchester, M13 9PL, UK

Abstract

We are developing methods to model transcription using mRNA expression (RNA-Seq) and RNA polymerase (pol-II ChIP-Seq) time course data. In our first application we model the motion of RNA polymerase during pre-mRNA elongation. We model the pol-II dynamics by using a spatio-temporal Gaussian process to described changes in pol-II density profiles across sites of the transcribed region [1]. We apply our model to infer the elongation speed and promoter-proximal pol-II activity for early targets of estrogen receptor in MCF7 breast cancer cells. Bayesian methods are used to infer the model parameters and associate our parameter estimates with levels of confidence. By clustering the inferred promoter-proximal pol-II activity profiles we can associate early-activated target genes with specific transcription factor binding patterns.

In our second application we link the pol-II dynamics with mRNA production and degradation in the same system using a simple linear differential equation. We again represent the pol-II dynamics as a Gaussian process and are able to exactly compute the data likelihood by exploiting the fact that a linear operation on a Gaussian process remains a Gaussian process. We find that for a certain number of target genes it is necessary to include an RNA-processing delay to get a reasonable fit to the data. We use Bayesian inference to infer the delay parameter and identify genes with strong evidence of a significant delay, about 11% of the genes where the signal is strong enough to fit the model. This delay appears to be related to splicing: we find that short genes tend to exhibit longer splicing-associated delay and there is also a positive association with genes that have a relatively long final intron.

Reference

  1. wa Maina C, Matarese F, Grote K, Stunnenberg HG, Reid G, Honkela A, Lawrence N, Rattray M (2014) Inference of RNA Polymerase II Transcription Dynamics from Chromatin Immunoprecipitation Time Course Data. PLoS Comp. Biol. 10(5), e1003598.