Prediction of nucleosome positioning based on transcription factor binding sites.

Xianfu Yi,Zhisong He,Weiren Cui,Yu-Dong Cai,Xiangyin Kong,Xiaoyu Zhang

doi:10.1371/journal.pone.0012495

Abstract

BackgroundThe DNA of all eukaryotic organisms is packaged into nucleosomes, the basic repeating units of chromatin. The nucleosome consists of a histone octamer around which a DNA core is wrapped and the linker histone H1, which is associated with linker DNA. By altering the accessibility of DNA sequences, the nucleosome has profound effects on all DNA-dependent processes. Understanding the factors that influence nucleosome positioning is of great importance for the study of genomic control mechanisms. Transcription factors (TFs) have been suggested to play a role in nucleosome positioning in vivo.Principal FindingsHere, the minimum redundancy maximum relevance (mRMR) feature selection algorithm, the nearest neighbor algorithm (NNA), and the incremental feature selection (IFS) method were used to identify the most important TFs that either favor or inhibit nucleosome positioning by analyzing the numbers of transcription factor binding sites (TFBSs) in 53,021 nucleosomal DNA sequences and 50,299 linker DNA sequences. A total of nine important families of TFs were extracted from 35 families, and the overall prediction accuracy was 87.4% as evaluated by the jackknife cross-validation test.ConclusionsOur results are consistent with the notion that TFs are more likely to bind linker DNA sequences than the sequences in the nucleosomes. In addition, our results imply that there may be some TFs that are important for nucleosome positioning but that play an insignificant role in discriminating nucleosome-forming DNA sequences from nucleosome-inhibiting DNA sequences. The hypothesis that TFs play a role in nucleosome positioning is, thus, confirmed by the results of this study.

Highlights

Of eukaryotic genomic DNA, 75–90% is wrapped around regularly spaced protein complexes called nucleosomes [1,2,3] (Figure 1), the fundamental building blocks of chromosomes
Our results are consistent with the notion that Transcription factors (TFs) are more likely to bind linker DNA sequences than the sequences in the nucleosomes
Our results imply that there may be some TFs that are important for nucleosome positioning but that play an insignificant role in discriminating nucleosome-forming DNA sequences from nucleosomeinhibiting DNA sequences

Summary

Introduction

Of eukaryotic genomic DNA, 75–90% is wrapped around regularly spaced protein complexes called nucleosomes [1,2,3] (Figure 1), the fundamental building blocks of chromosomes. The linker histone, H1, is associated with linker DNA and with the nucleosome core particle itself [7,8]. The length of linker DNA varies between species and cell types, as well as during differentiation and gene activation [7,8,9]. It is approximately 18 bp in Saccharomyces cerevisiae [7,8,9] and approximately 38 bp in humans [10]. The nucleosome consists of a histone octamer around which a DNA core is wrapped and the linker histone H1, which is associated with linker DNA. Transcription factors (TFs) have been suggested to play a role in nucleosome positioning in vivo

Objectives

Methods

Results

Conclusion