Inferring gene-regulatory networks using epigenomic priors

Area of Science:

• Genomics
• Epigenetics
• Systems Biology

Background:

• Gene regulatory networks (GRNs) control cellular functions.
• Accurate GRN inference is crucial for understanding gene regulation.
• Epigenomic data offers potential for improving GRN inference.

Purpose of the Study:

• To enhance the accuracy of in-silico GRN structure inference.
• To develop and validate a methodology using epigenomic prior networks.
• To explore the utility of different epigenomic data types for prior network construction.

Main Methods:

• Developed a computational methodology integrating epigenomic prior networks into GRN inference.
• Re-analyzed diverse epigenomic datasets (scRNA-seq, DNA methylation, chromatin accessibility, histone modifications) from 12 studies.
• Compared the effectiveness of DNA methylation and chromatin accessibility data for building epigenomic prior networks.

Main Results:

• The proposed methodology significantly improved in-silico GRN inference accuracy.
• DNA methylation data proved highly effective for inferring epigenomic prior networks, mirroring known structures.
• Inferring prior networks from DNA methylation data identified approximately eight times more candidate TF cis-regulations compared to chromatin accessibility data.
• Application to human embryonic development and breast cancer risk datasets revealed biologically relevant differential cis-regulation patterns.

Conclusions:

• Epigenomic prior networks substantially enhance GRN inference accuracy.
• DNA methylation data is a powerful resource for constructing epigenomic prior networks.
• The methodology provides a robust framework for generating hypotheses about gene regulation and epigenomic changes in development and disease.