A universal framework for detecting cis-regulatory diversity in DNA regions

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• High-throughput sequencing assays measure genome-wide gene regulatory activities like transcription factor (TF)-DNA binding and chromatin accessibility.
• Understanding sequence motifs and their combinations is crucial for explaining these activities.
• Existing methods often fail to account for the combinatorial diversity of regulatory elements.

Purpose of the Study:

• To introduce cisDIVERSITY, a novel statistical framework for analyzing regulatory regions.
• To model regulatory regions as diverse modules defined by combinations of motifs.
• To simultaneously learn motifs and account for combinatorial diversity without prior knowledge.

Main Methods:

• Developed a statistical framework, cisDIVERSITY, to model regulatory regions as diverse modules.
• Applied cisDIVERSITY to various high-throughput assay data, including GRO-cap, STARR-seq, and ATAC-seq.
• Utilized cisDIVERSITY for analyzing protein-DNA binding data and single-cell ATAC-seq data.

Main Results:

• cisDIVERSITY discovers distinct regulatory modules and TF binding site combinations from diverse assays.
• Identified potential cofactors for profiled TFs using protein-DNA binding data.
• Revealed tissue-specific regulatory modules from ATAC-seq data.
• Showcased cisDIVERSITY's ability to analyze dynamic changes in regulatory regions in single-cell ATAC-seq data, linking open chromatin states to future cell states.

Conclusions:

• cisDIVERSITY provides a generalizable statistical framework for dissecting combinatorial diversity in gene regulation.
• The method effectively identifies assay-specific and tissue-specific regulatory modules.
• cisDIVERSITY offers insights into the dynamic nature of regulatory elements and their role in cell state transitions.