A Grid Search-Based Multilayer Dynamic Ensemble System to Identify DNA N4-Methylcytosine Using Deep Learning Approach

Area of Science:

• Epigenetics and Molecular Biology
• Computational Biology and Bioinformatics
• Genomics

Background:

• DNA N4-methylcytosine (4mC) is a critical epigenetic modification influencing gene function, gene expression, and cellular processes.
• Precisely locating 4mC sites and their chromosomal distribution is essential for understanding its regulatory mechanisms.
• Current methods require enhancement for efficient and high-throughput 4mC site prediction.

Purpose of the Study:

• To develop an efficient and high-throughput intelligent computational system for predicting 4mC sites.
• To leverage natural language processing (word2vec) and deep learning (1D CNN) for 4mC prediction.
• To propose and evaluate a novel grid search-based multi-layer dynamic ensemble system (GS-MLDS).

Main Methods:

• Utilized the word2vec natural language processing method for feature extraction.
• Employed a multi-configured 1D Convolutional Neural Network (1D CNN) architecture.
• Developed a grid search-based multi-layer dynamic ensemble system (GS-MLDS) with optimized layer weighting.

Main Results:

• The GS-MLDS model achieved high prediction accuracies across eight benchmark datasets, with results ranging from 0.944 to 0.980.
• The proposed model demonstrated superior performance compared to 16 other distinct computational models.
• The system effectively enhances existing knowledge through its multi-layer ensemble approach.

Conclusions:

• The GS-MLDS system provides an accurate and efficient method for predicting 4mC sites.
• This computational approach significantly contributes to the understanding of DNA methylation and its role in gene regulation.
• The developed system offers a valuable tool for epigenetic research and analysis.