Related Experiment Video
Updated: Sep 16, 2025
Detection of Human Leukocyte Antigen Biomarkers in Breast Cancer Utilizing Label-free Biosensor Technology
Published on: March 24, 2015
Biomarker discovery for early breast cancer diagnosis using machine learning on transcriptomic data for biosensor
Kalaumari Mayoral-Peña1, Omar Israel González Peña2, Natalie Artzi3
1School of Engineering and Sciences, Campus Queretaro, Tecnologico de Monterrey, Queretaro, 76130, Mexico; Department of Medicine, Division of Engineering in Medicine, Brigham and Women's, Hospital Harvard Medical School, Boston, MA, 02115, USA.
This study developed a bioinformatics pipeline using machine learning to find genetic biomarkers for breast cancer classification. These biomarkers show potential for early detection and improved biosensor diagnostics, especially in low-income regions.
Area of Science:
- Bioinformatics and Computational Biology
- Genomics and Molecular Biology
- Oncology and Cancer Research
Background:
- Breast cancer is a leading cause of death globally, necessitating advanced diagnostic tools for early detection.
- Biosensors utilizing reliable biomarkers are crucial, particularly for resource-limited settings.
- Accurate classification of breast cancer subtypes (non-malignant, non-triple-negative, triple-negative) is vital for effective treatment.
Purpose of the Study:
- To identify novel genetic biomarkers for classifying breast cancer subtypes using a bioinformatics pipeline.
- To evaluate the performance of different gene selection approaches and machine learning algorithms.
- To assess the predictive power of identified biomarkers for patient survival and relapse.
Main Methods:
- Developed a novel bioinformatics pipeline integrating machine learning algorithms (MLAs) and five Gene Selection Approaches (GSAs).
- Employed LASSO, Membrane LASSO, Surfaceome LASSO, Network Analysis, and Feature Importance Score (FIS) for gene selection.
- Utilized Recursive Feature Elimination (RFE) and Genetic Algorithms (GAs) to reduce gene sets while maintaining high classification performance (F1 Macro ≥80%).
Main Results:
- Achieved high classification performance (F1 Macro or Accuracy 70.3%–97.2%) using selected gene sets.
- Identified 13 genes with significant predictive capabilities for five-year survival and 4 genes for relapse-free survival.
- Found overlap between identified genes and those in commercial diagnostic panels, validating their clinical relevance.
Conclusions:
- The novel bioinformatics pipeline effectively identifies robust genetic biomarkers for breast cancer classification.
- The identified biomarkers demonstrate potential for developing advanced biosensors for early breast cancer diagnosis and treatment monitoring.
- This approach offers a promising strategy for improving breast cancer diagnostics, especially in underserved regions.
