DNA Methylation Biomarker Discovery for Colorectal Cancer Diagnosis Assistance Through Integrated Analysis

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Colorectal cancer (CRC) diagnosis relies on accurate biomarkers.
• Identifying reliable biomarkers from tissue and blood samples remains a challenge.
• DNA methylation patterns offer potential for novel CRC biomarker discovery.

Purpose of the Study:

• To identify novel colorectal cancer (CRC) biomarkers with high classification accuracy.
• To leverage DNA methylation profiles and gene functional annotations for biomarker selection.
• To validate biomarker candidates using machine learning models.

Main Methods:

• Integrated analysis of CRC DNA methylation data from The Cancer Genome Atlas.
• Clustering of biomarker candidates based on promoter region proximity and functional annotations.
• Application of three machine learning techniques for model construction and performance evaluation.

Main Results:

• Ten genes exhibited significant methylation differences in both tissue and blood samples.
• A three-gene combination (ADHFE1, ADAMTS5, MIR129-2) demonstrated superior classification performance.
• The optimal biomarker combination achieved a Matthews correlation coefficient > 0.85 and an F1-score of 0.9.

Conclusions:

• Identified three robust CRC biomarkers through integrated DNA methylation analysis.
• These biomarkers show potential for a clinical diagnostic toolkit for CRC.
• The identified biomarkers warrant further investigation via liquid biopsies for clinical utility.