Related Experiment Video
Updated: Oct 24, 2025
Author Spotlight: Unveiling Transmembrane Protein Family-Related Markers in Gastric Cancer and Implications for Targeted Therapies
Published on: September 15, 2023
mRBioM: An Algorithm for the Identification of Potential mRNA Biomarkers From Complete Transcriptomic Profiles of
Changlong Dong1,2,3, Nini Rao1,2,3, Wenju Du1,2,3
1Center for Informational Biology, School of Life Sciences and Technology, University of Electronic Science and Technology of China, Chengdu, China.
Purpose:
In this work, an algorithm named mRBioM was developed for the identification of potential mRNA biomarkers (PmBs) from complete transcriptomic RNA profiles of gastric adenocarcinoma (GA).
Methods:
mRBioM initially extracts differentially expressed (DE) RNAs (mRNAs, miRNAs, and lncRNAs). Next, mRBioM calculates the total information amount of each DE mRNA based on the coexpression network, including three types of RNAs and the protein-protein interaction network encoded by DE mRNAs. Finally, PmBs were identified according to the variation trend of total information amount of all DE mRNAs. Four PmB-based classifiers without learning and with learning were designed to discriminate the sample types to confirm the reliability of PmBs identified by mRBioM. PmB-based survival analysis was performed. Finally, three other cancer datasets were used to confirm the generalization ability of mRBioM.
Results:
mRBioM identified 55 PmBs (41 upregulated and 14 downregulated) related to GA. The list included thirteen PmBs that have been verified as biomarkers or potential therapeutic targets of gastric cancer, and some PmBs were newly identified. Most PmBs were primarily enriched in the pathways closely related to the occurrence and development of gastric cancer. Cancer-related factors without learning achieved sensitivity, specificity, and accuracy of 0.90, 1, and 0.90, respectively, in the classification of the GA and control samples. Average accuracy, sensitivity, and specificity of the three classifiers with machine learning ranged within 0.94-0.98, 0.94-0.97, and 0.97-1, respectively. The prognostic risk score model constructed by 4 PmBs was able to correctly and significantly (∗∗∗ p < 0.001) classify 269 GA patients into the high-risk (n = 134) and low-risk (n = 135) groups. GA equivalent classification performance was achieved using the complete transcriptomic RNA profiles of colon adenocarcinoma, lung adenocarcinoma, and hepatocellular carcinoma using PmBs identified by mRBioM.
Conclusions:
GA-related PmBs have high specificity and sensitivity and strong prognostic risk prediction. MRBioM has also good generalization. These PmBs may have good application prospects for early diagnosis of GA and may help to elucidate the mechanism governing the occurrence and development of GA. Additionally, mRBioM is expected to be applied for the identification of other cancer-related biomarkers.
