Identification of hypertension gene expression biomarkers based on the DeepGCFS algorithm.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Hypertension is a major global health concern and a leading cause of cardiovascular mortality.
• Identifying reliable hypertension biomarkers from gene expression data is challenging due to small sample sizes and high dimensionality.
• Understanding hypertension mechanisms is crucial for effective prevention and treatment strategies.

Purpose of the Study:

• To develop a novel algorithm for identifying biologically significant hypertension gene biomarkers.
• To overcome the limitations of small sample sizes and high dimensionality in gene expression data analysis.
• To enhance the accuracy and reliability of hypertension biomarker discovery.

Main Methods:

• Proposed a deep graph clustering feature selection (DeepGCFS) algorithm.
• Utilized graph networks and Graph Neural Networks (GNNs) to model gene interactions.
• Employed link prediction, self-supervised learning, hybrid clustering, and integrated feature selection.

Main Results:

• The DeepGCFS algorithm identified ten significantly differentiated hypertension biomarkers.
• Achieved a classification performance AUC of 97.50%, outperforming existing methods.
• Validated findings on a separate dataset (GSE113439) with an AUC of 95.45% and identified seven significant genes.
• Six previously reported hypertension-associated genes were among the identified biomarkers.
• Demonstrated superior gene feature vector clustering performance compared to other methods.

Conclusions:

• The DeepGCFS algorithm effectively identifies biologically significant hypertension gene biomarkers.
• The proposed method offers significant advantages for hypertension biomarker discovery and analysis.
• This approach holds promise for advancing hypertension research and clinical applications.