CCPred: Global and population-specific colorectal cancer prediction and metagenomic biomarker identification at different molecular levels using machine learning techniques

Burcu Bakir-Gungor ¹, Mustafa Temiz ², Yasin Inal ¹

⁰Department of Computer Engineering, Faculty of Engineering, Abdullah Gul University, Kayseri, 38080, Turkey.

Computers in Biology and Medicine +

|

September 18, 2024

View abstract on PubMed

Summary

This study uses machine learning to analyze gut microbiome data for colorectal cancer (CRC) detection. It identifies key microbial species, enzymes, and pathways for improved CRC prediction and biomarker discovery.

Area Of Science

Microbiome research
Computational biology
Cancer genomics

Background

Colorectal cancer (CRC) is a leading cause of cancer death globally.
The gut microbiota plays a crucial role in CRC development and progression.
Accurate metagenomic biomarkers are needed for CRC diagnosis and treatment.

Purpose Of The Study

To evaluate CRC-associated metagenomic data at species, enzyme, and pathway levels.
To develop and assess machine learning models for CRC prediction using microbiome data.
To identify robust metagenomic biomarkers for CRC detection.

Main Methods

Utilized human gut microbiome sequencing data and relative abundance values.
Employed feature selection methods (SelectKBest, Information Gain, XGBoost) for global analysis.
Conducted population-specific analyses (within-population, LODO, cross-population) using four classification algorithms.

Main Results

Random Forest achieved high AUC values globally: 0.83 for species, 0.78 for enzymes, and 0.76 for pathways.
Identified potential taxonomic biomarker: *ruthenibacterium lactatiformanas*.
Identified potential enzyme and pathway biomarkers: RNA 2' 3' cyclic 3' phosphodiesterase and pyruvate fermentation to acetone pathway.

Conclusions

Machine learning models trained on metagenomic data show significant potential for enhanced CRC prediction.
Metagenomic analysis at species, enzyme, and pathway levels can improve biomarker discovery for CRC.
This study provides a framework for utilizing microbiome data in precision oncology for colorectal cancer.

Keywords:

Biomarkers Colorectal cancer Enzyme Machine learning Metagenomic Microbiome Pathway Species