Related Experiment Video
Updated: Sep 29, 2025
06:52
Discovery of Driver Genes in Colorectal HT29-derived Cancer Stem-Like Tumorspheres
Published on: July 22, 2020
6.7K
Machine Learning-Based Identification of Colon Cancer Candidate Diagnostics Genes
Saraswati Koppad1, Annappa Basava1, Katrina Nash2
1Department of Computer Science and Engineering, National Institute of Technology Karnataka, Mangalore 575025, India.
Biology
|March 26, 2022
Summary
This study identifies 34 key genes for diagnosing colorectal cancer (CRC) using machine learning. These findings offer a promising new diagnostic panel for early CRC detection and improved patient outcomes.
Area of Science:
- Genomics and Bioinformatics
- Cancer Research
- Machine Learning Applications
Background:
- Colorectal cancer (CRC) is a leading cause of cancer death globally, with high mortality due to challenges in early detection and understanding molecular mechanisms.
- Accurate diagnostic biomarkers are crucial for improving CRC outcomes, necessitating advanced analytical approaches for complex genomic data.
- High-throughput sequencing generates large datasets, driving the need for machine learning (ML) algorithms in identifying relevant molecular markers for CRC.
Purpose of the Study:
- To develop a novel machine learning-based experimental design for identifying diagnostic gene signatures in colorectal cancer.
- To leverage gene expression and clinical datasets to pinpoint genes with diagnostic potential for CRC.
- To explore the utility of ML algorithms in analyzing high-dimensional genomic data for biomarker discovery in cancer.
Main Methods:
- Employed six distinct machine learning classifiers (Adaboost, ExtraTrees, logistic regression, naïve Bayes, random forest, XGBoost) to analyze gene expression and clinical data.
- Evaluated classifier performance using metrics such as accuracy, sensitivity, specificity, F1 score, and AUROC across various dataset combinations.
- Conducted gene ontology enrichment analysis on identified differentially expressed genes (DEGs) and mapped them with microRNAs (miRNAs).
Main Results:
- Random Forest models demonstrated superior performance compared to other evaluated machine learning methods.
- Identified a panel of 34 differentially expressed genes with high diagnostic accuracy for colorectal cancer.
- Pathway and gene set enrichment analyses were performed on the 34 genes, revealing potential miRNA-hub interactions.
Conclusions:
- Successfully identified 34 genes with high accuracy, forming a potential diagnostic panel for colorectal cancer.
- The developed ML-based approach effectively pinpointed key genes for CRC diagnosis.
- These findings pave the way for more accurate and earlier detection of colorectal cancer through novel biomarkers.
