An Integrated Approach for Identifying Molecular Subtypes in Human Colon Cancer Using Gene Expression Data

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Colorectal cancer (CRC) molecular subtypes are crucial for personalized treatment.
• Existing gene expression-based subtype identification methods for CRC lack consistency.
• A robust and progressive methodological approach is needed for accurate subtype discovery.

Purpose of the Study:

• To develop and validate a novel, integrated approach for identifying molecular subtypes of colon cancer.
• To improve the accuracy and efficiency of colon cancer subtyping using gene expression data.
• To establish a reliable method for discovering finer, clinically relevant colon cancer subtypes.

Main Methods:

• Integration of Bayesian robust principal component analysis (BRPCA) for denoising.
• Application of directed bubble hierarchical tree (DBHT) algorithm for hierarchical clustering.
• Utilizing an improved differential evolution based feature selection (DEFS_W) algorithm for gene selection, prioritizing sample imbalance.

Main Results:

• Identification of 7 distinct molecular subtypes within the colon cancer cohort.
• Discovery of 44 key feature genes differentiating these subtypes.
• Demonstrated superior performance in identifying finer subtypes with fewer genes compared to previous studies.
• Validated the approach using the Cancer Genome Atlas (TCGA) mRNA gene expression dataset (153 colon cancer, 19 normal samples).

Conclusions:

• The proposed integrated methodology effectively identifies molecular subtypes of colon cancer.
• This approach offers a more refined classification of colon cancer subtypes with a reduced gene set.
• The generic methodology holds potential for application in identifying subtypes of other cancer types.