4mCpred-EL

4mCpred-EL predicts DNA N^4-methylcytosine (4mC) sites in the mouse genome to support analysis of 4mC distribution relevant to epigenetic regulation.

Key Features:

• Machine Learning Integration: Incorporates four distinct machine learning algorithms to generate initial predictions for 4mC sites.
• Feature Encoding Diversity: Utilizes seven different feature encodings to represent sequence information for model training.
• Ensemble Learning Framework: Uses probabilistic outputs from initial models as feature vectors in an ensemble framework to combine model strengths.
• Performance Metrics: Reports an accuracy of 79.80% and a Matthews Correlation Coefficient (MCC) of 0.591, outperforming seven other classifiers by 1.5%–5.9% in accuracy and 3.2%–11.7% in MCC.

Scientific Applications:

• Epigenetic Research: Enables identification of 4mC site distribution to investigate epigenetic mechanisms of gene regulation and cellular processes.
• Genomic Studies: Supports species-specific analyses of N4-methylcytosine in the mouse genome for comparative and functional genomics.
• Pre-screening for Experimental Validation: Provides candidate 4mC site predictions to facilitate targeted experimental validation.

Methodology:

Genomic sequences are input and preprocessed; seven feature encodings are extracted; four different ML algorithms generate probabilistic predictions; those probabilistic outputs are used as feature vectors in an ensemble learning integration.