---

🧬 Transcriptomic Biomarker Discovery in Breast Cancer

Overview

This project performs an integrated bioinformatics analysis to identify potential breast cancer biomarkers using gene expression data.

The workflow combines differential expression analysis, functional enrichment, protein interaction networks, and survival analysis to prioritize clinically relevant biomarker candidates.

---

Dataset

GEO Accession: GSE42568 Organism: Homo sapiens

Dataset:

• 54,675 genes
• 121 breast tissue samples

Analysis was performed using R and Bioconductor packages.

---

Workflow

GEO Dataset
     ↓
Differential Expression Analysis (limma)
     ↓
GO & KEGG Enrichment
     ↓
PPI Network Analysis (STRING)
     ↓
Survival Analysis
     ↓
Final Biomarker Candidates

---

Methods

Differential Expression Analysis

Tool: limma

Criteria:

Adjusted P-value < 0.05
|log2FC| > 1

Results:

Category	Genes
Upregulated	2731
Downregulated	2970

---

Functional Enrichment

GO Analysis

Top biological processes:

• Cell-substrate adhesion
• Actin filament organization
• Cell-matrix adhesion
• PI3K-AKT signaling regulation

KEGG Pathways

Top pathways:

• Focal adhesion
• Integrin signaling
• ECM-receptor interaction
• Proteoglycans in cancer

---

Biomarker Identification

Candidates were selected using:

✔ Differential expression ✔ PPI network connectivity ✔ Survival association

Final biomarkers:

Gene	Degree	Cox P-value
CDC20	7	6.33e-06
AURKA	8	6.09e-04
FOXA1	4	4.97e-04
GATA3	4	2.80e-03
SOX4	2	5.14e-03

---

Repository Structure

├── main.R
├── scripts/
├── results/
│   ├── figures/
│   └── tables/
├── README.md
└── .gitignore

---

Running the Pipeline

source("main.R")

All results will be generated in the results/ directory.

---

Future Improvements

• Machine learning based biomarker prediction
• ROC validation
• External dataset validation
• Multi-omics integration

---