Pathway analysis for RNA-Seq data using a score-based approach.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Pathway/gene-set approaches analyze high-level biological phenomena from gene expression data.
• Resampling methods are preferred over independence-assuming approaches for gene expression analysis.
• Existing methods are often not optimized for RNA-Seq data's discrete counts and potential outliers.

Purpose of the Study:

• To develop and validate a pathway testing approach specifically for RNA-Seq data.
• To improve statistical power and handle normalization factors in RNA-Seq pathway analysis.
• To provide a robust platform for pathway analysis with appropriate error control.

Main Methods:

• Data transformation techniques applied to RNA-Seq counts to address discrete nature and library size variations.
• Utilized null approximations to quadratic form statistics for self-contained and competitive pathway testing.
• Developed an integrated platform for RNA-Seq pathway analysis, avoiding computationally intensive permutations.

Main Results:

• The proposed method demonstrates appropriate Type I error control without permutation.
• The approach shows enhanced statistical power compared to competing methods in various settings.
• Pathway analysis on rat and human cell line data confirmed biological interpretability of findings.

Conclusions:

• The developed RNA-Seq pathway testing approach is powerful and provides reliable results.
• The method offers a practical and effective tool for analyzing gene expression data from RNA-Seq experiments.
• This approach enhances the biological insights obtainable from transcriptomic studies.