SENIES

SENIES predicts enhancers and their regulatory strength by applying deep learning to integrate DNA shape information with sequence-derived features.

Key Features:

• Two-Layer Architecture: A first layer discriminates enhancers from non-enhancers and a second layer predicts the strength of identified enhancers.
• Integration of DNA Shape Information: Incorporates DNA shape features that capture 3D structural characteristics relevant to transcription factor binding preferences.
• Sequence-Derived Features: Utilizes one-hot encoding and k-mer representations as input feature sets.
• Deep Learning Models: Employs deep learning techniques for feature learning and prediction.
• Ensemble Classifier: Integrates multiple feature-derived predictors using an ensemble classifier approach.

Scientific Applications:

• Enhancer identification: Distinguishing enhancers from non-enhancer genomic sequences.
• Enhancer strength prediction: Predicting regulatory strength to inform studies of gene regulation.
• Regulatory network analysis: Supporting elucidation of regulatory networks that govern gene expression.
• Biomedical research: Aiding studies of genetic disorders, developmental biology, and personalized medicine through improved enhancer characterization.

Methodology:

The method uses a deep learning-based two-layer predictor that combines DNA shape features with one-hot encoding and k-mer representations and applies an ensemble classifier to integrate multiple feature-derived predictors.