Motif-based models accurately predict cell type-specific distal regulatory elements

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Understanding how DNA sequence dictates cell-specific gene regulation is crucial.
• Distal cis-regulatory elements control gene expression but their sequence code is complex.

Purpose of the Study:

• To present Bag-of-Motifs (BOM), a novel computational framework for predicting cell-type-specific enhancers.
• To demonstrate BOM's accuracy and interpretability across diverse species.

Main Methods:

• Representing cis-regulatory elements as unordered counts of transcription factor (TF) motifs.
• Utilizing gradient-boosted trees for predictive modeling.
• Validating predictions through experimental construction of synthetic enhancers.

Main Results:

• BOM accurately predicts cell-type-specific enhancers in mouse, human, zebrafish, and Arabidopsis.
• BOM outperforms complex deep-learning models in prediction accuracy with fewer parameters.
• Experimental validation confirmed that predicted motif sets drive cell-type-specific expression.

Conclusions:

• BOM provides a scalable and interpretable framework for dissecting cis-regulatory grammar.
• A simple sequence code within distal regulatory regions drives cell-type-specific gene activity.
• BOM offers broad applicability for gene regulation studies across species.