Identification of Cell Cycle-Regulated Genes by Convolutional Neural Network

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Cell cycle-regulated genes exhibit periodic expression, crucial for understanding cell cycle processes.
• Challenges in cell cycle-regulated gene detection include high false positives and low overlap between methods.
• Accurate identification of these genes is vital for advancing cell cycle research.

Purpose of the Study:

• To introduce DLGene, a novel computational framework for enhanced cell cycle-regulated gene detection.
• To address limitations of existing methods in identifying cell cycle-regulated genes and their expression patterns.
• To analyze the biological functions of identified cell cycle gene subtypes.

Main Methods:

• Developed DLGene, a deep learning framework utilizing convolutional neural networks.
• Transformed gene expression data into categorical states to reveal distinct expression patterns.
• Compared DLGene's performance against six traditional machine learning algorithms.

Main Results:

• DLGene demonstrated superior and balanced sensitivity and specificity compared to other machine learning methods.
• Identified four distinct subtypes of cell cycle-regulated genes based on their expression patterns.
• Provided novel insights into cell cycle mechanisms through functional analysis of representative genes.

Conclusions:

• DLGene offers a robust approach for accurate cell cycle-regulated gene detection and expression pattern analysis.
• The framework successfully distinguishes cell cycle genes and their subtypes, improving upon existing methods.
• Functional analysis of identified gene subtypes offers new perspectives on cell cycle regulation.