Identification of sample-specific regulations using integrative network level analysis.

Area of Science:

• Computational biology and bioinformatics
• Cancer genomics and systems biology

Background:

• Tumor heterogeneity poses challenges for genome-wide data analysis.
• Existing network methods often require large datasets, limiting their application.
• A need exists for network methods capable of single-sample analysis to address heterogeneity and sample size limitations.

Purpose of the Study:

• To introduce Differentially Expressed Regulation Analysis (DERA), a novel network method.
• To enable analysis of gene regulatory networks at a single sample level.
• To identify molecular functions and subgroups within cancer samples.

Main Methods:

• Developed DERA, a network method integrating gene expression data with biological network information.
• Constructed sample-specific networks to identify shared or subgroup-specific regulations.
• Applied DERA to triple-negative breast cancer (TNBC) and high-grade serous ovarian cancer (HGS-OvCa) datasets.

Main Results:

• DERA identified 110 core regulations and 28 subnetworks in TNBC, linked to oncogenic activity, proliferation, survival, invasion, and metastasis.
• Discovered 31 regulations specific to TNBC, aiding in understanding this subtype.
• Identified common regulations between TNBC and HGS-OvCa, and demonstrated superior reproducibility and sensitivity compared to GSEA and SPIA on small datasets.

Conclusions:

• DERA is a novel method for identifying similarly active subnetworks within sample groups.
• Application to breast and ovarian cancer data confirmed DERA's ability to identify reliable and significant regulations with high reproducibility.
• An R package for DERA is publicly available.