ConGEMs: Condensed Gene Co-Expression Module Discovery Through Rule-Based Clustering and Its Application to...

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Cancer Research

Background:

• Transcriptomic analysis generates vast genomic data, particularly for cancer research.
• Traditional methods often compare cancer versus control groups for individual genes.
• Association rule mining offers a powerful approach to uncover causal relationships between transcripts.

Purpose of the Study:

• To introduce novel rule-based similarity measures: weighted rank-based Jaccard and Cosine.
• To propose a computational framework (ConGEMs) for detecting condensed gene co-expression modules.
• To explore biomarker modules from transcriptomic data using association rule learning.

Main Methods:

• Identification of differentially expressed genes using an empirical Bayes test.
• Application of the RANWAR algorithm to determine association rules from gene expression data.
• Computation of integrated similarity scores using novel weighted similarity measures for clustering co-expressed rule-modules.

Main Results:

• Development of a novel framework for detecting condensed gene co-expression modules (ConGEMs).
• Identification of condensed markers supported by literature, KEGG pathways, and Gene Ontology annotations.
• Demonstrated superior performance of the proposed method over traditional gene-module discovery measures on lung squamous cell carcinoma and cervical carcinogenesis datasets.

Conclusions:

• The proposed rule-based method effectively identifies condensed gene co-expression modules from transcriptomic data.
• This approach facilitates the exploration of potential biomarker modules in cancer research.
• The framework offers a valuable tool for uncovering complex gene relationships and causal effects.