MaGIC: a machine learning tool set and web application for monoallelic gene inference from chromatin

Area of Science:

• Epigenetics
• Genomics
• Computational Biology

Background:

• A significant proportion of human and mouse autosomal genes exhibit random monoallelic expression (MAE).
• MAE is an epigenetic mechanism involving allele-specific gene expression that varies across clonal cell lineages.
• MAE is cell-type specific, and its mapping is crucial for understanding its biological function, yet detection remains challenging.

Purpose of the Study:

• To present a user-friendly, open-source software pipeline for inferring monoallelic gene expression from chromatin data.
• To implement a previously reported method that uses sequence-independent chromatin signatures for MAE detection.
• To facilitate the mapping of MAE across diverse cell types.

Main Methods:

• Development of the MaGIC (monoallelic gene inference from chromatin) software pipeline.
• Utilizing ChIP-seq data to identify a sequence-independent chromatin signature associated with MAE.
• Providing a user-friendly interface and open-source code for accessibility.

Main Results:

• The MaGIC pipeline successfully implements a sensitive and specific method for identifying genes subject to MAE.
• The pipeline leverages readily available ChIP-seq data for MAE detection.
• The source code and a Shiny app are publicly available.

Conclusions:

• The MaGIC pipeline empowers researchers to map monoallelic expression in various cell types.
• The tool supports the use of existing models and the training of new models with additional chromatin marks.
• This facilitates broader investigation into the biological roles of MAE.