Identification of key biomarkers for STAD using filter feature selection approaches

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Gastric cancer (GC) is a leading cause of cancer death globally.
• Early detection through diagnostic biomarkers is crucial for improving patient outcomes.
• High-dimensional biological data analysis presents challenges in biomarker discovery.

Purpose of the Study:

• To identify a gene signature for distinguishing gastric tumor and normal samples.
• To validate the diagnostic performance of the identified gene signature in independent cohorts.
• To highlight the utility of feature selection methods in biomarker discovery for gastric cancer.

Main Methods:

• Utilized the limma method combined with joint mutual information (JMI), a machine learning algorithm.
• Identified an 11-gene signature from a stomach adenocarcinoma cohort.
• Validated the signature's classifying performance on two additional GC datasets.

Main Results:

• An 11-gene signature demonstrated strong performance in distinguishing tumor from normal gastric samples.
• Several identified candidate genes showed correlations with GC tumor progression and patient survival.
• The study successfully validated the gene signature in independent GC datasets.

Conclusions:

• Feature selection approaches, like JMI, are effective for analyzing high-dimensional biological data.
• The identified 11-gene signature shows promise as a diagnostic biomarker for gastric cancer.
• This approach can improve accuracy and reduce workload in identifying potential tumor biomarkers.