Deep-4mCGP: A Deep Learning Approach to Predict 4mC Sites in Geobacter pickeringii by Using Correlation-Based Feature...

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• 4-methylcytosine (4mC) is a DNA modification influencing key biological processes like DNA replication and gene expression.
• Accurate identification of 4mC sites is crucial for understanding their hereditary functions.

Purpose of the Study:

• To develop a robust deep learning model for recognizing 4mC sites specifically in the bacterium Geobacter pickeringii.

Main Methods:

• DNA sequences from Geobacter pickeringii were encoded using binary and k-mer composition feature descriptors.
• Feature fusion was followed by optimization using correlation and a Gradient-Boosting Decision Tree (GBDT) algorithm with Incremental Feature Selection (IFS).
• Optimized features were input into a 1D Convolutional Neural Network (CNN) for 4mC site classification.

Main Results:

• The developed deep learning model achieved an accuracy of 0.868 on independent test data.
• This performance represents a 4.2% improvement compared to existing models for 4mC site prediction.

Conclusions:

• The proposed deep learning approach effectively identifies 4mC sites in Geobacter pickeringii.
• This method offers a more accurate tool for studying the functional roles of 4mC modifications in bacterial genomes.