Label propagation-based semi-supervised feature selection on decoding clinical phenotypes with RNA-seq data.

Area of Science:

• Computational biology and bioinformatics
• Neuroscience and neurodegenerative disease research
• Genomics and transcriptomics

Background:

• Neurodegenerative diseases impact behavior, cognition, and mental functions.
• The molecular pathogenesis of these complex diseases remains unclear.
• Identifying stage-specific biomarkers is crucial for understanding disease progression.

Purpose of the Study:

• To develop a novel computational approach for prioritizing disease-associated genes.
• To identify key genes and molecular mechanisms underlying different clinical phenotypes in neurodegenerative diseases.
• To decode changes in behavioral and mental characteristics during disease progression.

Main Methods:

• Developed a label propagation-based semi-supervised feature selection (LPFS) framework.
• Integrated label propagation clustering with feature selection using gene expression profiles.
• Performed Gene Ontology (GO) and KEGG pathway enrichment analysis for functional insights.

Main Results:

• LPFS demonstrated superior performance compared to state-of-the-art methods on Huntington's disease data.
• Identified affected pathways including TGF-beta signaling, immune, and inflammatory responses.
• Highlighted significant expression changes in genes such as SLC4A11, ZFP474, and AMBP.

Conclusions:

• The LPFS model effectively selects key genes for different disease phenotypes.
• Experiments with Huntington's disease data revealed involvement of astrocytes, microglia, and GABAergic neurons.
• The study provides insights into the molecular mechanisms of neurodegenerative disease progression.