Pro-Coffee

Pro-Coffee aligns homologous promoter regions using a dinucleotide substitution matrix derived from TRANSFAC functional binding-site alignments to improve identification of orthologous regulatory sequences and transcription factor binding sites.

Key Features:

• Dinucleotide Substitution Matrix: Pro-Coffee employs a dinucleotide substitution matrix estimated from alignments of functional binding sites sourced from the TRANSFAC database.
• Tailored for Regulatory Sequences: The method is specialized for promoter and regulatory region alignment, distinguishing it from general-purpose aligners.
• Validation Framework: The method was evaluated on a dataset comprising several thousand families of orthologous promoters to assess ortholog versus paralog identification.
• Enhanced Ortholog Identification Accuracy: Pro-Coffee achieved 80.4% accuracy in identifying true orthologs compared with averages of 73.5% and 77.6% for other methods.
• Multi-species ChIP-seq Validation: Validation included a procedure using multi-species ChIP-seq data to test alignment of experimentally detected binding sites across species.
• Binding Site Alignment Performance: Pro-Coffee correctly aligned 331 transcription factor binding sites, approximately 16.5% more than the default-method average (284) and outperforming trained methods (316).
• Correlation with Ortholog Classification: A strong correlation was observed between a method's ortholog classification accuracy and its proficiency in aligning proven binding sites, indicating that training on the ortholog dataset increases functional informativeness.

Scientific Applications:

• Gene regulation studies: Generate more accurate promoter-region alignments to support analysis of transcription factor binding and regulatory mechanisms.
• Comparative genomics: Identify orthologous promoters across species to investigate evolutionary conservation and divergence in regulatory networks.

Methodology:

Pro-Coffee uses a dinucleotide substitution matrix estimated from alignments of functional binding sites from TRANSFAC and was evaluated on several thousand families of orthologous promoters, with additional validation using multi-species ChIP-seq comparing trained and untrained methods and metrics based on ortholog classification accuracy and counts of correctly aligned binding sites.