A unified computational framework for modeling genome-wide nucleosome landscape

Area of Science:

• Molecular Biology
• Genomics
• Computational Biology

Background:

• Nucleosomes are fundamental to eukaryotic chromatin structure.
• The influence of DNA sequence on nucleosome distribution is a long-standing debate.
• Understanding these principles is crucial for deciphering genome organization.

Purpose of the Study:

• To quantify the impact of DNA sequence features on nucleosome occupancy and positioning.
• To develop a computational framework for predicting genome-wide nucleosome maps.
• To differentiate the roles of G+C content, nucleotide periodicity, and poly(dA:dT) tracts.

Main Methods:

• Developed a computational framework to learn nucleosome number and positioning energy.
• Analyzed genome-wide nucleosome maps from Saccharomyces cerevisiae.
• Compared in vitro and in vivo nucleosome predictions.

Main Results:

• G+C content is a primary determinant of MNase-derived nucleosome occupancy, but with potential biases.
• Poly(dA:dT) tracts consistently deter nucleosome formation.
• 10.5 bp nucleotide periodicity aids rotational, not translational, positioning.
• DNA sequence alone can partially explain nucleosome arrays near transcription start sites.
• Genomic sequence largely explains nucleosome occupancy at transcription termination sites.

Conclusions:

• DNA sequence features significantly shape the genome-wide nucleosome landscape.
• The computational model accurately predicts both in vitro and in vivo nucleosome maps.
• Experimental methods like MNase-seq can introduce biases affecting occupancy interpretation.