International Environmental 'Omics Synthesis Conference, University of Cardiff, 9-11 September 2013

Description

Possible binding, or functionally important binding? A case study of human transcription factor p53 and some suggestions for the future

A transcription factor may be able to bind a stretch genomic DNA but, in itself, this does not imply important biological consequences. I present a recent, computational approach for predicting binding sites for the transcription factor p53 within the human genome (Lim et al. 2013). On the assumption that a sequence must be accessible for any sort of binding to actually occur, our model combines both sequence and histone modification data to predict binding sites. When assessed by ability to predict binding sites known from chromatin immunoprecipitation (ChIP) experiments, the performance of the combined sequence­‐plus­‐modification model is similar to that of a simpler model that ignores histone modification data. Despite this negative result, the combined­‐evidence model is much better at predicting functionally important binding sites, defined as sites near genes induced by p53. We propose that this is because sequence and ChIP data reflect primarily biophysical properties of protein­‐DNA interactions, whereas chromatin modification data capture biologically important functional information. I discuss implications for detecting transcription factor binding sites in general, and for the controversy over the ENCODE project.

Reference.

Lim, J­‐H, Iggo, R and Barker, D (2013) Models incorporating chromatin modification data identify functionally important p53 binding sites. Nucleic Acids Research 41. 5582­‐5593.