Improved cancer biomarkers identification using network-constrained infinite latent feature selection

Area of Science:

• Bioinformatics
• Computational Biology
• Cancer Genomics

Background:

• Identifying cancer biomarkers is crucial for targeted therapies.
• Current methods for gene expression analysis have limitations in identifying significant cancer-related genes.

Purpose of the Study:

• To develop a novel method for identifying reliable and effective cancer biomarkers.
• To improve the accuracy of cancer diagnosis and the biological significance of identified biomarkers.

Main Methods:

• Proposed a novel method combining infinite latent feature selection (ILFS) with functional interaction (FI) networks.
• Applied the network-constrained ILFS (NCILFS) method to gene expression data from five cancer types.
• Utilized gene ontology (GO) biological process (BP) enrichment analysis and gene set enrichment analysis (GSEA).

Main Results:

• NCILFS demonstrated improved diagnostic prediction compared to original ILFS and other methods.
• NCILFS identified a greater number of known oncogenes.
• Functional enrichment analysis revealed more biologically significant gene sets identified by NCILFS.

Conclusions:

• Network-constrained ILFS (NCILFS) is a powerful tool for identifying cancer-related genes.
• The method offers higher discriminative power and biological significance for biomarker discovery.
• NCILFS advances the field of bioinformatics for cancer research and targeted therapy development.