SeqEnhDL

SeqEnhDL classifies cell type-specific enhancers from DNA sequence using deep learning to identify sequence features that distinguish enhancers across cell types.

Key Features:

• Deep Learning Models: Implements Multi-Layer Perceptron (MLP), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) architectures trained on sequential k-mer features (k = 5, 7, 9, and 11).
• Enhancer Classification: Classifies "strong enhancer" chromatin states across nine ENCODE cell types by distinguishing these enhancers from non-coding sequences using k-mer fold changes relative to randomly selected non-coding regions.
• Performance Superiority: Outperforms gkm-SVM and DanQ and can directly discriminate enhancers between different cell types.

Scientific Applications:

• Cell Type-Specific Enhancer Identification: Enables identification of enhancers that drive gene regulation in specific cell types based on sequence features.
• Tissue-Specificity Analysis: Supports analysis of tissue- and cell type-specific regulatory elements to inform studies of development and disease mechanisms.

Methodology:

Uses sequential k-mer features (k = 5, 7, 9, 11) computed as fold changes of k-mers relative to randomly selected non-coding sequences as input to MLP, CNN, and RNN models.