Using machine learning to identify gene interaction networks associated with breast cancer

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• Breast cancer (BC) is a prevalent malignancy with unclear etiology.
• Obesity is a known risk factor, implicating obesity-related genes in BC development.
• Understanding gene interactions, not just single genes, is crucial for elucidating BC's complex genetic mechanisms.

Purpose of the Study:

• To construct a gene interaction network for breast cancer (BC) using SNP data.
• To identify potential pathogenic genes and their interactions involved in BC.
• To gain insights into the genetic mechanisms underlying BC progression.

Main Methods:

• Utilized joint density-based non-parametric differential interaction network analysis and classification (JDINAC) to build the BC gene interaction network.
• Analyzed single nucleotide polymorphisms (SNPs) in 953 BC patients and 963 controls.
• Validated hub genes and interactions using TCGA RNA-seq data and UK Biobank datasets, followed by GO and KEGG enrichment analysis.

Main Results:

• Identified a BC gene interaction network, highlighting Body Mass Index (BMI) and menopause as significant risk factors.
• LEP, LEPR, XRCC6, and RETN were identified as key hub genes.
• LEPR polymorphisms (rs1137101 and rs4655555) showed significant association with BC; enriched genes are involved in energy regulation and fat signaling pathways.

Conclusions:

• Gene interaction networks derived from SNP data offer novel insights into breast cancer (BC) pathogenesis.
• This study identifies potential therapeutic targets and enhances understanding of BC's genetic underpinnings.