Ranking genome-wide correlation measurements improves microarray and RNA-seq based global and targeted co-expression...

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Co-expression networks are crucial for inferring gene associations and predicting gene functions.
• Pathway Level Coexpression (PLC) analyzes transcriptional landscapes of specific pathways using guide genes.
• Defining gene co-expression is a critical step in network construction.

Purpose of the Study:

• To compare the performance of different distance measures for gene co-expression network construction.
• To evaluate these measures on both global networks and PLCs using diverse datasets.
• To identify the most effective method for analyzing plant biological pathways.

Main Methods:

• Evaluated Pearson Correlation Coefficient (PCC), Spearman Correlation Coefficient (SCC), Highest Reciprocal Rank (HRR), Mutual Information (MI), and Partial Correlations (PC).
• Applied methods to global co-expression networks and Pathway Level Coexpression (PLC) analyses.
• Utilized microarray and RNA-seq datasets from Arabidopsis thaliana across five specific pathways.

Main Results:

• PCC ranked with HRR demonstrated superior performance compared to other distance measures.
• This combination proved effective for both global network construction and PLC.
• The method showed particular strength in clustering genes into biologically relevant subpathways.

Conclusions:

• Pearson Correlation Coefficient (PCC) combined with Highest Reciprocal Rank (HRR) is the recommended approach for gene co-expression network analysis.
• This method is well-suited for both global networks and Pathway Level Coexpression (PLC) using microarray and RNA-seq data.
• It enhances the ability to identify biologically meaningful gene groupings within pathways.