Discovering Gene Regulatory Elements Using Coverage-Based Heuristics.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Combining data mining and sequencing methods (RNA-seq, ChIP-seq) aims to find genomic regulatory motifs linked to phenotypes.
• Current motif discovery algorithms generate lengthy lists of potential transcription factor binding sites, complicating experimental validation.

Purpose of the Study:

• To address the challenge of selecting relevant motifs from extensive lists.
• To introduce and solve the motif selection problem using novel algorithms.

Main Methods:

• Development of coverage-based search heuristics for motif selection.
• Analysis of 203 ChIP-seq experiments from the ENCyclopedia of DNA Elements (ENCODE) project.

Main Results:

• The proposed algorithms generate motifs with high sensitivity and specificity.
• The greedy algorithm demonstrated superior performance, selecting a median of two motifs per group with 77% median sensitivity.
• New insights into the human genome's regulatory code were revealed.

Conclusions:

• The developed motif selection algorithms effectively reduce the complexity of identifying phenotype-related regulatory elements.
• These findings enhance our understanding of genomic regulation and facilitate experimental validation of transcription factor binding sites.