Targeted co-expression networks for the study of traits

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Weighted Gene Co-expression Network Analysis (WGCNA) is a common method for gene co-expression network generation.
• WGCNA often produces large, functionally complex modules that are difficult to interpret.
• There is a need for more precise and biologically relevant gene network analysis tools.

Purpose of the Study:

• To introduce Targeted Gene Co-expression Networks (TGCN) as a novel method for generating more focused gene co-expression modules.
• To compare the performance of TGCN with WGCNA in terms of module precision and biological interpretability.
• To demonstrate the utility of TGCN in identifying molecular pathways related to specific biological traits, using Alzheimer's disease as a case study.

Main Methods:

• TGCN refines LASSO regression to identify key transcripts that predict a trait of interest based on gene expression.
• Co-expression modules are constructed around these predictive transcripts.
• Algorithm properties were validated using gene expression data from 13 brain regions (Genotype-Tissue Expression project).
• An APP-TGCN was created using The Religious Orders Study and Memory and Aging Project dataset to investigate APP's role in Alzheimer's disease.

Main Results:

• TGCN generates more precise gene co-expression modules compared to WGCNA.
• These TGCN modules exhibit more specific yet biologically rich functional annotations.
• The APP-TGCN successfully identified molecular pathways associated with APP in Alzheimer's disease.
• Key findings were validated in two independent cohorts.

Conclusions:

• TGCN provides a new framework for creating smaller, biologically relevant gene networks.
• This method is advantageous for high-throughput, hypothesis-driven research.
• TGCN enhances the ability to decipher complex biological relationships and identify disease-associated pathways.