Differential transcript usage analysis incorporating quantification uncertainty via compositional measurement error...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Differential transcript usage (DTU) analysis identifies changes in relative transcript expression between conditions.
• Existing DTU methods face challenges with speed and scalability for large sample sizes.
• Quantification uncertainty in RNA-seq expression estimates is often overlooked by current DTU approaches.

Purpose of the Study:

• To develop a computationally efficient and scalable method for DTU detection.
• To incorporate quantification uncertainty into DTU analysis.
• To improve sensitivity and reduce false positives in DTU detection.

Main Methods:

• Proposed CompDTU method utilizes compositional regression to model transcript proportions.
• Leverages fast matrix-based computations for enhanced speed and scalability.
• Extended CompDTUme method incorporates quantification uncertainty from RNA-seq data.

Main Results:

• CompDTU demonstrates excellent sensitivity and reduced false positives compared to existing methods.
• CompDTUme further improves performance, particularly for genes with high quantification uncertainty and sufficient sample size.
• Applied to breast cancer data, new methods showed reduced computation time and identified novel DTU genes across subtypes.

Conclusions:

• CompDTU and CompDTUme provide efficient, scalable, and accurate solutions for DTU analysis.
• The methods effectively handle large datasets and address quantification uncertainty.
• These advancements facilitate the discovery of novel DTU events in complex biological systems like cancer.