Comparison of Normalization Methods for Analysis of TempO-Seq Targeted RNA Sequencing Data

Area of Science:

• Genomics
• Transcriptomics
• Bioinformatics

Background:

• Bulk RNA sequencing (RNA-Seq) offers genome-wide transcription insights.
• Targeted RNA sequencing, like TempO-Seq, provides a resource-efficient alternative for transcriptomic analysis.
• Effective data normalization is crucial for accurate interpretation of RNA-Seq data, yet methods for TempO-Seq remain under-investigated.

Purpose of the Study:

• To evaluate the performance of various normalization methods on TempO-Seq data.
• To identify the most effective normalization strategy for accurate gene expression analysis in two-class comparisons using TempO-Seq.

Main Methods:

• Simulated RNA-Seq count data from human HepaRG cells using TempO-Seq platform.
• Applied seven different normalization methods to the simulated data.
• Assessed normalization performance based on fold-change accuracy, specificity, sensitivity, and clustering agreement using limma and K-means.

Main Results:

• Upper Quartile (UQ) normalization demonstrated superior performance in maintaining fold-change levels and exhibited high specificity and sensitivity.
• K-means clustering analysis showed the best agreement with fold-change assignments when using UQ normalization.
• DESeq2, Counts Per Million (CPM), and Total Counts (TCs) normalization methods also provided reliable results for fold-changes greater than or equal to 2.0.

Conclusions:

• UQ normalization is recommended as the optimal method for analyzing two-class comparisons of TempO-Seq data.
• CPM, TCs, and DESeq2 normalization are viable alternatives for TempO-Seq data with absolute fold-changes of 2.0 or higher.
• These findings provide essential guidance for researchers to improve the reliability of TempO-Seq gene expression data analysis.