Identification of active transcriptional regulatory elements from GRO-seq data

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Global run-on and sequencing (GRO-seq) modifications can identify active transcriptional regulatory elements (TREs).
• Cap-based enrichment is often required for accurate TRE identification using GRO-seq.
• Existing methods may not fully capture the functional landscape of TREs.

Purpose of the Study:

• Introduce discriminative regulatory-element detection from GRO-seq (dREG), a novel machine learning method.
• Enable identification of active TREs from standard GRO-seq data without cap-based enrichment.
• Facilitate simultaneous assessment of TREs, gene expression, and other transcriptional features.

Main Methods:

• Developed dREG, a machine learning approach utilizing support vector regression.
• Applied dREG to analyze standard GRO-seq data.
• Validated predicted TREs against established marks of transcriptional activation.

Main Results:

• dREG accurately identifies active TREs from GRO-seq data without cap enrichment.
• Predicted TREs show higher enrichment for expression quantitative trait loci, disease polymorphisms, H3K27ac, and transcription factor binding compared to alternative assays.
• Surveyed TREs across eight human cell types, revealing global patterns of TRE function.

Conclusions:

• dREG provides a sensitive and accurate method for identifying active TREs using readily available GRO-seq data.
• This approach enhances the study of transcriptional regulation and its functional elements.
• The findings offer new insights into the global distribution and function of TREs in human cells.