A stochastic model for identifying differential gene pair co-expression patterns in prostate cancer progression

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Identifying gene co-expression patterns across cancer stages is crucial for understanding carcinogenesis.
• Existing methods lack statistical rigor for assessing gene co-expression changes during cancer progression.
• A specific algorithm is needed to pinpoint gene pairs correlated with cancer advancement.

Purpose of the Study:

• To develop and validate a novel analytical method for identifying differentially co-expressed gene pairs during cancer progression.
• To assess the correlation of identified gene pairs with cancer progression and their enrichment in disease-specific pathways.

Main Methods:

• Developed the Stochastic process model for Identifying differentially co-expressed Gene pairs (SIG method).
• Applied the SIG method to prostate cancer datasets (hormone sensitive vs. resistant, healthy vs. cancerous).
• Compared SIG method results with existing statistical methods using progression analysis, gene pair identification effectiveness, and pathway enrichment analysis.

Main Results:

• The SIG method identified 428,582 and 303,992 gene pairs in the two prostate cancer datasets, respectively.
• Gene pairs identified by SIG demonstrated high correlation with cancer progression (large PS, TPR) and pathway enrichment (large PES).
• SIG method outperformed existing methods in progression analysis (small RS) and pathway enrichment.

Conclusions:

• The SIG method reliably identifies cancer progression-correlated gene pairs.
• The SIG method excels in gene pair ontology and pathway enrichment analyses.
• This approach offers an effective means to understand carcinogenesis by tracking cancer progression.