Related Experiment Video
Updated: Jun 26, 2026
09:40
Characterization of Functionally Associated miRNAs in Glioblastoma and their Engineering into Artificial Clusters for Gene Therapy
Published on: October 4, 2019
Glioma identification from microRNA biomarkers using machine learning.
Rakesh Kanth Andugala1, Alyson Cieslik2, Maria Braoudaki2
1School of Physics, Engineering and Computer Science, University of Hertfordshire, Hatfield, United Kingdom.
Frontiers in Systems Biology
|June 25, 2026
Summary
Machine learning accurately identifies gliomas using microRNA (miRNA) expression data. This approach aids in early detection and discovers novel miRNA biomarkers for brain tumor classification.
Area of Science:
- Biomarkers and Diagnostics
- Computational Biology and Bioinformatics
- Oncology and Cancer Research
Background:
- Gliomas are aggressive brain tumors, with traditional diagnostics being invasive and costly.
- MicroRNAs (miRNAs) are key regulators of gene expression, and their dysregulation is implicated in cancer development.
- miRNAs in bodily fluids offer a promising avenue for minimally invasive glioma detection and classification.
Purpose of the Study:
- To develop and evaluate a machine learning (ML) pipeline for glioma identification using miRNA expression data.
- To investigate potential miRNA biomarkers for distinguishing gliomas from controls and meningiomas.
- To compare ML model performance using various feature selection and classification algorithms.
Main Methods:
- Utilized miRNA expression data from four Gene Expression Omnibus (GEO) datasets.
- Applied five feature selection techniques (LASSO, mRMR, ReliefF, RFE, RF importance).
- Employed six ML algorithms (LR, KNN, DT, RF, SVM, XGB) with and without SMOTE oversampling, assessed via 5-fold cross-validation.
Main Results:
- Achieved up to 100% accuracy in binary classification (glioma vs. controls).
- Reached up to 100% F1-score in multi-class classification (glioma vs. meningioma vs. controls) using KNN and XGB classifiers.
- Identified seven potential miRNA biomarkers (miR-125a-3p, miR-4276, miR-4648, miR-4763-3p, miR-663a, miR-6784-5p, miR-873-3p), validated on an independent dataset.
Conclusions:
- The developed ML pipeline effectively classifies gliomas using miRNA expression data.
- The identified miRNAs represent promising novel biomarkers for glioma diagnosis and classification.
- This minimally invasive approach holds potential for improved early detection and personalized treatment of brain tumors.
