PhyME: a probabilistic algorithm for finding motifs in sets of orthologous sequences

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Identifying transcription factor binding sites (TFBS) is crucial for understanding gene regulation.
• Heterogeneous sequence data, including orthologous sequences across species, presents challenges for accurate TFBS discovery.

Purpose of the Study:

• To develop a novel algorithm for discovering transcription factor binding sites.
• To integrate sequence overrepresentation and cross-species conservation into a unified probabilistic score for motif significance.

Main Methods:

• The algorithm employs the Expectation-Maximization (EM) technique.
• It accommodates user-defined phylogenetic trees to model evolutionary relationships between orthologous sequences.
• The method is designed to scale efficiently with the number of species and sequence length.

Main Results:

• The proposed algorithm successfully integrates overrepresentation and cross-species conservation.
• Evaluations on synthetic data and real datasets from yeast, fly, and human demonstrate its efficacy.
• The approach shows improved performance in motif discovery compared to methods not utilizing cross-species information.

Conclusions:

• The developed algorithm enhances transcription factor binding site discovery.
• Exploiting information from multiple species significantly improves motif identification accuracy.
• This approach offers a robust tool for analyzing regulatory sequences in comparative genomics.