High heterogeneity undermines generalization of differential expression results in RNA-Seq analysis

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• RNA sequencing (RNA-Seq) is vital for monitoring transcriptome changes in oncology.
• Tumor heterogeneity can cause high variation in gene expression, impacting the reproducibility of differential expression (DE) results.
• The reproducibility of DE results and the reasons for non-reproducible differentially expressed genes (DEGs) remain understudied.

Purpose of the Study:

• Investigate the reproducibility of DE results across varying biological replicate numbers (3-24).
• Identify the causes of poor reproducibility in DEGs detected by RNA-Seq.
• Assess the impact of tumor heterogeneity on gene expression analysis.

Main Methods:

• RNA sequencing (RNA-Seq) data analysis.
• Differential gene expression analysis.
• Statistical evaluation of reproducibility across different sample sizes.

Main Results:

• Poor reproducibility of DE results was observed even with large sample sizes.
• Many detected DEGs were sample-specific, not genuinely differentially expressed.
• High gene expression variation, driven by biological variability and outlier data, is the primary cause of poor reproducibility.

Conclusions:

• Significant tumor and normal sample heterogeneity undermines the generalizability of DE results.
• Larger sample sizes (>=10) are recommended for RNA-Seq experiments to mitigate biological variability.
• DE results require cautious interpretation and robust validation due to inherent heterogeneity.