Impact of RNA-seq data analysis algorithms on gene expression estimation and downstream prediction.

Area of Science:

• Bioinformatics
• Genomics
• Biostatistics

Background:

• Next-generation sequencing (NGS) technologies like RNA-sequencing (RNA-seq) are crucial for medical and health applications.
• Selecting appropriate RNA-seq analysis pipelines for biomarker identification presents a significant challenge.
• The US Food and Drug Administration (FDA) Sequencing Quality Control (SEQC) project investigated 278 RNA-seq analysis pipelines.

Purpose of the Study:

• To assess the impact of joint RNA-seq pipeline effects on gene expression estimation.
• To evaluate the influence of RNA-seq pipelines on the downstream prediction of disease outcomes.
• To develop metrics for guiding the selection of optimal RNA-seq pipelines.

Main Methods:

• Developed and applied three metrics: accuracy, precision, and reliability for gene expression estimation.
• Investigated the correlation between these metrics and downstream prediction performance.
• Utilized two cancer datasets: SEQC neuroblastoma and TCGA lung adenocarcinoma.

Main Results:

• RNA-seq pipeline components jointly and significantly affect gene expression estimation accuracy.
• This impact extends to the downstream prediction of cancer outcomes.
• Pipelines yielding more accurate, precise, and reliable gene expression data performed better in disease outcome prediction.

Conclusions:

• RNA-seq pipeline choice critically influences gene expression accuracy and disease prediction.
• The developed metrics can guide users in selecting pipelines for improved accuracy, precision, and reliability.
• Optimized RNA-seq analysis enhances gene expression-based disease outcome prediction.