MetaDiff

MetaDiff performs differential isoform expression analysis on RNA-Seq data by applying random-effects meta-regression to account for uncertainty in isoform expression estimates.

Key Features:

• Isoform-level analysis: Operates at isoform-level resolution to detect differential isoform expression from RNA-Seq data.
• Random-effects meta-regression model: Implements a random-effects meta-regression to model isoform-level expression and manage estimation uncertainties.
• Covariate adjustment: Supports adjustment for covariates to control confounding variables in differential expression analyses.
• Uncertainty handling: Models uncertainties arising from ambiguous reads and variability in precision across samples.
• Computational performance and validation: Demonstrated computational speed, robustness, increased statistical power, and false discovery rate control in simulations and on a human heart failure RNA-Seq dataset.
• Implementation: Implemented using Java and R.

Scientific Applications:

• Complex disease isoform analysis: Detecting isoform-level differential expression in studies of complex diseases.
• Gene regulation and disease mechanism studies: Investigating gene regulation mechanisms and the implications of isoform changes in health and disease.
• Application to specific cohorts: Analysis of RNA-Seq datasets such as human heart failure to identify differential isoform expression.

Methodology:

MetaDiff applies a random-effects meta-regression to isoform-level RNA-Seq expression estimates, explicitly modeling ambiguity from reads and sample-specific precision and allowing inclusion of covariates.