Exploring chromatin conformation and gene co-expression through graph embedding.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Gene co-expression and 3D genome organization are crucial for biological processes.
• High-throughput chromosome conformation capture (Hi-C) technologies provide insights into genome structure.
• Developing computational tools to link gene co-expression with chromatin conformation is a growing research area.

Purpose of the Study:

• To develop a computational framework for predicting gene co-expression networks from chromatin conformation data.
• To investigate the correlation between chromatin topology and gene co-expression patterns.
• To identify physical interaction patterns most predictive of gene co-expression.

Main Methods:

• Construction of a gene chromatin interaction network based on physical interaction profiles.
• Application of two graph embedding techniques to generate low-dimensional vector representations of genes.
• Training a classifier on gene embedding pairs to predict co-expression.

Main Results:

• Graph embedding techniques outperformed traditional methods relying on manual topological features.
• Random walk-based graph embedding demonstrated superior performance.
• Demonstrated a non-linear relationship between chromatin conformation and gene regulation.
• Gene topological embeddings encode valuable information for downstream analyses.

Conclusions:

• Computational framework effectively predicts co-expression networks from Hi-C data.
• Advanced graph embedding methods are essential for encoding complex chromatin information.
• Gene topological embeddings derived from chromatin data hold significant potential for understanding gene regulation.