i4mC-Mouse

i4mC-Mouse predicts N^4-methylcytosine (4mC) sites in the mouse genome using ensemble Random Forest models built from multiple DNA sequence encoding schemes.

Key Features:

• Multiple Encoding Schemes: Represents sequences using six encodings: K-space Nucleotide Composition (KSNC), k-mer Nucleotide Composition (Kmer), Mono Nucleotide Binary Encoding (MBE), Dinucleotide Binary Encoding, Electron-Ion Interaction Pseudo Potentials (EIIP), and Dinucleotide Physicochemical Composition.
• Random Forest-Based Models: Trains a Random Forest (RF)-based model for each individual encoding scheme.
• Model Integration: Produces final predictions by linear combination of the probability scores from the individual RF models.
• Performance Metrics: Achieves an independent-test accuracy of 0.816 and a Matthews Correlation Coefficient (MCC) of 0.633, and outperforms 4mCpred-EL by approximately 2.5% in accuracy and 5% in MCC.
• Feature Contribution: Reports relative contributions of encodings to predictive performance with Kmer, KSNC, MBE, and EIIP accounting for 10%, 45%, 25%, and 20%, respectively.

Scientific Applications:

• Epigenetic site identification: Enables genome-wide prediction of N^4-methylcytosine (4mC) sites in mouse DNA sequences to support mapping of DNA methylation patterns.
• Gene regulation and differentiation studies: Supports research into gene regulation mechanisms and cell differentiation processes influenced by 4mC modifications.

Methodology:

Sequences are encoded using six schemes (KSNC, Kmer, MBE, dinucleotide binary, EIIP, dinucleotide physicochemical), individual Random Forest models are trained per encoding, and final predictions are obtained by linear integration of the models' probability scores.