iEnhancer-DCSV

iEnhancer-DCSV predicts enhancers and their strength from DNA sequences to support studies of gene regulation.

Key Features:

• Data Encoding: DNA sequences are represented using one-hot encoding and nucleotide chemical property (NCP) features.
• Modified DenseNet: A modified densely connected convolutional network (DenseNet) extracts complex sequence features from the encoded inputs.
• Improved CBAM: An enhanced convolutional block attention module (CBAM) applies channel and spatial attention to prioritize significant features.
• Fully Connected Neural Network: Prioritized features are processed by a fully connected neural network to produce prediction probabilities for enhancer presence and strength.
• Ensemble Learning with Voting: Multiple models are combined via ensemble learning and a voting mechanism to determine final classifications.

Scientific Applications:

• Enhancer identification: Predicts the presence of enhancer regions in genomic sequences.
• Enhancer strength assessment: Predicts relative enhancer strength to inform studies of regulatory element activity.
• Genomic research and large-scale studies: Provides computational predictions of enhancers and their strength to aid investigations of gene regulation and related biological processes.

Methodology:

Sequences are encoded with one-hot and NCP; features are extracted with a modified DenseNet; CBAM performs channel and spatial attention evaluation; a fully connected neural network outputs prediction probabilities for enhancer presence and strength; ensemble learning with a voting mechanism produces final classifications.