Comparing the normalization methods for the differential analysis of Illumina high-throughput RNA-Seq data

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology Techniques
• Computational Biology

Background:

• RNA-Sequencing (RNA-Seq) is a crucial technology for quantifying gene expression.
• Effective normalization is vital for accurate RNA-Seq data analysis.
• A comparative analysis of normalization methods is needed to guide experimental design.

Purpose of the Study:

• To compare the performance of various RNA-Seq normalization methods.
• To identify the most suitable normalization approaches based on alignment accuracy and read length.
• To provide guidelines for selecting optimal normalization strategies in RNA-Seq experiments.

Main Methods:

• Evaluated eight non-abundance estimation methods (RC, UQ, Med, TMM, DESeq, Q, RPKM, ERPKM) and two abundance estimation methods (RSEM, Sailfish).
• Utilized real Illumina RNA-Seq data (35- and 76-nucleotide sequences) from the MAQC project and simulated reads.
• Assessed normalization performance using Spearman correlation with MAQC qRT-PCR values for 996 genes.

Main Results:

• Non-abundance methods (RC, UQ, Med, TMM, DESeq, Q) yielded similar results.
• RPKM showed high correlation for 35-nucleotide sequences but low correlation for 76-nucleotide sequences.
• Sailfish demonstrated superior performance over RSEM for 35-nucleotide sequences; RSEM performed comparably to no normalization for 76-nucleotide sequences.
• Adding a poly-A tail increased alignment numbers but did not enhance normalization results.

Conclusions:

• RC, UQ, Med, TMM, DESeq, and Q provided no significant improvement in gene expression normalization.
• Sailfish with RPKM was most effective under low alignment accuracy conditions.
• RC normalization is adequate for gene expression calculation when alignment accuracy is high.
• Poly-A tail inclusion should be disregarded in differential gene expression analysis.