Studying Nucleosomes Positioning by a Multi-Layer Model

Abstract

Eukaryotic DNA is packaged into a highly compact and dynamic structure called chromatin. While this packaging allows the cell to organize a large and complex genome in the nucleus, it can also block the access of transcription factors and other proteins to DNA. Nucleosomes are the fundamental repeating units of eukaryotic chromatin. Nucleosome position can be regulated in vivo by multi-subunit chromatin remodeling complexes, and their position can influence gene expression in eukaryotic cells. Alterations in chromatin structure, and hence in nucleosome organization, can result in a variety of diseases, including cancer, highlighting the need to achieve a better understanding of the molecular processes modulating chromatin dynamics. In this paper, we present a new method, Multi-Layer Model (MLM), that can be successfully used to identify genome wide nucleosome positions starting from microarray data. This new approach allows a significant reduction in computational time as well as a better structural view of the input data when compared to a Hidden Markov Model HMM recently used for the same purpose. In order to evaluate the accuracy of the method the $MLM$ has been validated on synthetic data and compared with the HMM on real microarray data of the Saccharomyces cerevisiae. By the way, although the MLM is faster than the other methods devoted to the same purpose when tested on real data, the processing of large microarray data (more than 10^4 base pairs) is a real case where the MLM becomes computationally demanding making essential the use of high performance computing architectures. To this purpose, two parallel implementations of the MLM are described and analysed.