Size-based expectation maximization for characterizing nucleosome positions and subtypes.

Area of Science:

• * Genomics
• * Molecular Biology
• * Bioinformatics

Background:

• * Nucleosome profiling commonly uses MNase-seq, assuming uniform 147 bp DNA protection.
• * Atypical nucleosomes with shorter DNA protection are overlooked by standard analysis packages.
• * Understanding diverse nucleosome types is crucial for deciphering gene regulation.

Purpose of the Study:

• * Introduce the size-based expectation maximization (SEM) package for nucleosome analysis.
• * Enable characterization of diverse nucleosome subtypes from MNase-seq data.
• * Overcome limitations of rigid assumptions in existing nucleosome-calling tools.

Main Methods:

• * Developed SEM, a package employing a hierarchical Gaussian mixture model.
• * Utilized SEM to estimate nucleosome positions and automatically identify subtypes based on DNA fragment length distributions.
• * Validated SEM performance against existing packages using benchmark datasets.

Main Results:

• * SEM achieves comparable accuracy to standard packages while uniquely identifying nucleosome subtypes.
• * Applied to MNase-H2B-ChIP-seq data, SEM identified short-fragment, canonical, and di-nucleosomes.
• * Characterized two subtypes of short-fragment nucleosomes based on chromatin accessibility, with accessible subtypes enriched at TSSs and CTCF sites.

Conclusions:

• * SEM provides a robust platform for exploring non-canonical nucleosome subtypes.
• * Accessible short-fragment nucleosomes, identified by SEM, are prevalent in regulatory regions and associate with chromatin remodelers.
• * This work enhances the ability to study nucleosome heterogeneity and its regulatory implications.