Gene-level differential analysis at transcript-level resolution.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression analysis is crucial for understanding biological processes.
• Traditional gene-level analysis using RNA-sequencing counts can obscure transcript-level variations.
• Existing methods may lack sensitivity or require complex quantification steps.

Purpose of the Study:

• To develop a more sensitive and accurate gene-level differential expression analysis method.
• To propose a novel approach that leverages transcript-level data.
• To enable efficient analysis using transcript compatibility counts.

Main Methods:

• Implementing an 'analysis first, aggregation second' strategy.
• Aggregating p-values from transcript-level differential analysis.
• Utilizing transcript compatibility counts from pseudoalignment to bypass quantification.
• Applying the method to gene ontology enrichment analysis.

Main Results:

• The proposed method significantly increases sensitivity and accuracy in gene-level differential expression analysis.
• The approach is compatible with fast and accurate pseudoalignment methods.
• The method demonstrates robustness across different biological levels, including gene ontologies.

Conclusions:

• Aggregating transcript p-values provides a superior approach to gene-level differential expression analysis.
• The proposed method offers a computationally efficient and accurate alternative for RNA-sequencing data.
• This technique enhances the biological insights obtainable from transcriptomic studies.