Vaccine and Infectious Disease Division

PING tracks down nucleosomes

Our DNA is packaged into our cells with many different methods of organization. At the smallest level, DNA is wound around protein complexes called nucleosomes; together, these proteins and DNA are known as chromatin. Although nucleosomes are somewhat routinely spaced along the DNA, their position depends on both the physical flexibility and the function of the underlying DNA sequence. Understanding a cell’s nucleosome positioning can shed light on how genes are activated and other aspects of chromatin. With the advent of full-genome sequencing came techniques to view an entire cell or organism’s nucleosome map. One common technique uses an enzyme, MNase, to chew away DNA that is between nucleosomes; the nucleosome-bound DNA is then sequenced. Another technique, chromatin immunoprecipitation or ChIP-seq, uses vibration to break non-nucleosomal DNA. ChIP-seq is a more straightforward method but often does not yield as high resolution data as MNase digestion.

VIDD associate member Dr. Raphael Gottardo and colleagues devised a computational model, PING, to predict nucleosome position based on experimental results from ChIP-seq data, with the goal of improving biological predictions from this less precise technique. The researchers used ChIP-seq data from yeast and mouse experiments to show that their model was able to accurately predict nucleosome positions in these organisms. To apply their analyses to biological predictions, they looked at the binding positions of two mouse transcription factors, Foxa2 and Pdx1. Nucleosome positioning data can be combined with genome-wide transcription factor binding data to get a more accurate picture of functional transcription factor binding sites, that is, those sites on the DNA that are not occluded by nucleosomes. The researchers’ analyses of ChIP-seq data revealed the same functional binding sites of these two transcription factors as had been previously published from MNase digestion data.

Zhang X, Robertson G, Woo S, Hoffman BG, Gottardo R. Probabilistic inference for nucleosome positioning with MNase-based or sonicated short-read data. PLoS One. 2012;7(2):e32095.