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

• Published in: BMC bioinformatics Vol. 20; no. 1; pp. 106 - 5
• Main Authors: Vinogradova, Svetlana, Saksena, Sachit D., Ward, Henry N., Vigneau, Sébastien, Gimelbrant, Alexander A.
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 28.02.2019; BioMed Central; BMC
• Subjects: Alleles; Animal models; Applications programs; Artificial intelligence; Biochemistry; Bioinformatics; Biology; Business metrics; Cancer; Cell culture; Chromatin; Chromatin signature; Data processing; Datasets; Epigenetic inheritance; Epigenetics; Gene expression; Generalized linear models; Genes; Genetic research; Genomes; House mouse; Inference; Internet software; Learning algorithms; Machine learning; Mammals; Mapping; Methods; Monoallelic expression; Open source software; Pipelines; Shiny app; Software; Software pipeline; Source code; Web services; Web site management software; United States; United States > US; Chromatin; Chromatin signature; Software pipeline; Monoallelic expression; Shiny app
• ISSN: 1471-2105; 1471-2105
• DOI: 10.1186/s12859-019-2679-7

A large fraction of human and mouse autosomal genes are subject to random monoallelic expression (MAE), an epigenetic mechanism characterized by allele-specific gene expression that varies between clonal cell lineages. MAE is highly cell-type specific and mapping it in a large number of cell and tissue types can provide insight into its biological function. Its detection, however, remains challenging. We previously reported that a sequence-independent chromatin signature identifies, with high sensitivity and specificity, genes subject to MAE in multiple tissue types using readily available ChIP-seq data. Here we present an implementation of this method as a user-friendly, open-source software pipeline for monoallelic gene inference from chromatin (MaGIC). The source code for the MaGIC pipeline and the Shiny app is available at https://github.com/gimelbrantlab/magic . The pipeline can be used by researchers to map monoallelic expression in a variety of cell types using existing models and to train new models with additional sets of chromatin marks.