Eponine

Eponine predicts transcription start sites (TSS) in mammalian genomes using probabilistic models that combine hybrid machine-learning and sequence analysis to identify promoter-associated signals.

Key Features:

• Machine Learning Integration: Uses a hybrid probabilistic machine-learning method to construct models of promoters for TSS detection.
• Specificity and Positional Accuracy: Demonstrates specificity greater than 70% and provides good positional accuracy in detecting transcription start sites.
• Coverage of Human TSS: Can build useful promoter models for over 50% of human transcription start sites.
• Motif and Sequence Signal Detection: Identifies sequence-based signals including a TATA box–like motif and flanking regions enriched in C-G nucleotides.

Scientific Applications:

• Gene Expression Regulation: Maps promoter regions and TSS to study mechanisms of gene expression regulation.
• Alternative Splicing: Investigates genes with multiple TSS to assess impacts on alternative splicing events.
• Genomic Annotation: Enhances genomic annotation by accurately mapping promoter regions and transcription initiation sites.

Methodology:

Applies a probabilistic approach that combines hybrid machine-learning with sequence analysis to model promoters and identify key motifs (TATA box–like signals and C-G–enriched flanking regions).