Predicting cell-type-specific gene expression from regions of open chromatin

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Cell-type-specific gene expression relies on transcription factor (TF) binding to regulatory DNA elements.
• Predicting this expression is challenging due to the difficulty in locating distal regulatory regions.

Purpose of the Study:

• To identify proximal and distal regulatory elements across 19 human cell types using DNase-seq data.
• To develop models for predicting cell-type-specific gene expression patterns.
• To assess the contribution of TF binding motifs in regulatory elements to gene expression prediction.

Main Methods:

• Utilized DNase-seq data from 19 human cell types to map open chromatin regions (regulatory elements).
• Associated DNase I hypersensitive sites (DHSs) with genes and analyzed promoter characteristics (CG dinucleotide content, DNA accessibility).
• Trained predictive classifiers incorporating TF sequence motif matches within DHSs.

Main Results:

• Identified significant differences in CG dinucleotide content and DNA accessibility between up-regulated, down-regulated, and constitutively expressed genes.
• TF motif matches in DHSs substantially improved gene expression prediction accuracy compared to using only proximal promoter sequences.
• Achieved competitive performance in distinguishing cell-type-specific up-regulated genes and genes with opposing regulation within the same cell type.
• Identified novel candidate cell-type-specific regulators and predicted their functions.

Conclusions:

• DNase-seq data effectively identifies regulatory elements for predicting mammalian gene expression.
• TF binding motifs within open chromatin regions are crucial for accurate prediction of cell-type-specific gene expression.
• The developed models offer testable predictions for TF functions and binding activities.