Related Experiment Video
Updated: Jun 8, 2026
09:16
Detection of a CDH1 Rare Transcript Variant in Fresh-frozen Gastric Cancer Tissues by Chip-based Digital PCR
Published on: February 5, 2018
GCBRGCN: Integration of ceRNA and RGCN to Identify Gastric Cancer Biomarkers
Peng Zhi1,2,3,4,5, Yue Liu6, Chenghui Zhao2,3,4,5
1Chinese PLA Medical School, Chinese PLA General Hospital, Beijing 100853, China.
Bioengineering (Basel, Switzerland)
|March 28, 2025
Summary
This study introduces a novel graph neural network model for gastric cancer (GC) biomarker discovery, integrating multiple RNA types for improved diagnostic and prognostic prediction. The model identified new potential GC biomarkers, enhancing oncological research.
Area of Science:
- Oncology
- Bioinformatics
- Genomics
Background:
- Gastric cancer (GC) is a significant global health challenge.
- Current GC biomarker discovery often relies on single RNA types, limiting predictive power.
- Integrating multiple RNA interactions may enhance biomarker identification for GC diagnosis and prognosis.
Purpose of the Study:
- To develop a novel computational framework for identifying gastric cancer biomarkers.
- To leverage the competing endogenous RNA (ceRNA) network and whole transcriptomics data for biomarker discovery.
- To improve the accuracy of gastric cancer biomarker prediction using advanced machine learning.
Main Methods:
- Developed the Gastric Cancer Biomarker Relation Graph Convolutional Neural Network (GCBRGCN) model.
- Integrated competing endogenous RNA (ceRNA) networks with clinical and transcriptomics data.
- Utilized relational graph convolutional networks (RGCN) for biomarker prediction.
Main Results:
- The GCBRGCN model achieved an AUC of 0.8172 in predicting GC biomarkers, outperforming traditional methods.
- Identified three novel potential GC biomarkers: CCNG1, CYP1B1, and CITED2.
- Characterized FOXC1 and LINC00324 as significant biomarkers for both GC prognosis and diagnosis.
Conclusions:
- The study presents a novel framework for gastric cancer biomarker identification using multi-RNA interactions.
- The GCBRGCN model demonstrates high efficacy in predicting GC biomarkers.
- Highlights the importance of exploring complex RNA interactions in oncological research.
