Enhancer-MDLF

Enhancer-MDLF identifies cell-specific enhancers using a multi-input deep learning framework to improve enhancer prediction accuracy and reveal transcription factor binding site motifs associated with regulatory activity.

Key Features:

• Multi-Input Deep Learning Framework: Uses a multi-input deep learning architecture to integrate multiple data inputs and improve enhancer identification accuracy, addressing limitations of previous methods such as Enhancer-IF.
• Superior Performance Across Datasets: Demonstrates improved performance across eight distinct human cell lines on both generic enhancer datasets and enhancer-promoter interaction datasets.
• Transfer Learning for Specificity Prediction: Employs transfer learning to enhance generalization across different cell types for predicting enhancer specificity.
• Model Interpretation for Regulatory Insights: Provides model interpretation capabilities to identify transcription factor binding site motifs associated with enhancer regions.

Scientific Applications:

• Cell-specific enhancer identification: Enables accurate identification of cell-specific enhancers to study gene regulation and genomic regulatory networks.
• Enhancer-promoter interaction analysis: Applicable to enhancer-promoter interaction datasets for investigating regulatory interactions.
• Regulatory motif discovery: Facilitates identification of transcription factor binding motifs linked to enhancer function and regulatory mechanisms.

Methodology:

Multi-input deep learning architecture integrating multiple data inputs, application of transfer learning for cross–cell-type generalization, and model interpretation techniques to identify transcription factor binding site motifs.