CSMET: comparative genomic motif detection via multi-resolution phylogenetic shadowing

Area of Science:

• Genomics
• Evolutionary Biology
• Bioinformatics

Background:

• Transcription factor binding sites (TFBSs) undergo functional turnover (gain/loss) during evolution.
• Conventional methods struggle to capture TFBS evolutionary dynamics, especially in complex genomes.
• Comparative genomic motif discovery for non-conserved sites is challenging.

Purpose of the Study:

• To develop a novel computational method for detecting functional turnover of TFBSs.
• To improve the discovery of non-conserved motifs in comparative genomics.
• To provide a principled approach for analyzing lineage-specific TFBS evolution.

Main Methods:

• Conditional Shadowing via Multi-resolution Evolutionary Trees (CSMET), a context-dependent probabilistic graphical model.
• Modeling functional specifications of aligned sequence segments over lineages.
• Integration with a hidden Markov model for autocorrelated evolutionary rates.

Main Results:

• CSMET effectively models TFBS gain and loss across taxa.
• The method provides a computable posterior distribution of motifs under TFBS turnover.
• CSMET outperforms state-of-the-art comparative genomic motif finders on simulated and real data.

Conclusions:

• CSMET offers a principled and effective approach for comparative genomic motif discovery.
• The method accurately accounts for lineage-specific TFBS evolution.
• CSMET advances the study of cis-regulatory element evolution and function.