consensusDE: an R package for assessing consensus of multiple RNA-seq algorithms with RUV correction.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• No single RNA-seq algorithm consistently outperforms others for differential expression (DE) analysis.
• Removal of Unwanted Variation (RUV) is a technique to stabilize DE results.
• Integrating RUV with multiple DE algorithms presents a computational challenge.

Purpose of the Study:

• To develop and evaluate consensusDE, a tool for automating DE analysis by integrating multiple RNA-seq algorithms.
• To assess the impact of RUV on DE stability and accuracy across different algorithms.
• To provide guidelines for applying RUV within a consensus DE framework.

Main Methods:

• consensusDE was developed to automate the identification of significant DE genes.
• The tool integrates results from multiple RNA-seq algorithms with optional RUV integration.
• It requires minimal user input, accepting sample group tables and BAM files or count tables.

Main Results:

• RUV increased fold change stability between algorithms.
• RUV improved False Discovery Rate (FDR) in low replication settings for intersecting DE genes.
• The effect of RUV was algorithm-specific and diminished with increased replication, emphasizing the importance of sample size.

Conclusions:

• consensusDE facilitates robust DE analysis by combining multiple algorithms and RUV.
• RUV's utility in consensus DE depends on the specific algorithm and replication level.
• The study provides practical considerations for applying RUV in consensus DE settings.