dSreg

dSreg infers regulators of alternative splicing (AS) by integrating RNA-seq-derived exon inclusion estimates with RNA-binding protein (RBP) binding-site features using a Bayesian model.

Key Features:

• Bayesian integration model: Employs a Bayesian model that integrates RNA-seq data with RBP binding-site features.
• Regulator identification and activity quantification: Identifies key RBP regulators of AS changes and quantifies their regulatory activity.
• Exon inclusion estimation: Simultaneously estimates changes in exon inclusion rates.
• Low-coverage sensitivity: Improves sensitivity and specificity for detecting AS changes, including at low RNA-seq read coverage, as demonstrated on simulated data.
• Benchmarking against enrichment methods: Shows improved performance over over-representation analysis and gene set enrichment analysis on knock-down RBP datasets such as ENCODE.
• Large-scale integration: Integrates large amounts of low-coverage RNA-seq data for AS analysis.
• Implementation: Implemented in Python using Stan for statistical modeling.

Scientific Applications:

• Regulator discovery: Identification of RNA-binding proteins that regulate alternative splicing changes.
• Activity profiling: Quantification of RBP regulatory activity across conditions.
• Knock-down experiment analysis: Analysis of knock-down RBP experiments, including datasets from ENCODE.
• Low-coverage and simulation studies: Detection and validation of AS changes in low-coverage RNA-seq and simulated data.

Methodology:

Bayesian model integrating RNA-seq exon inclusion estimates with RBP binding-site features, with simultaneous estimation of exon inclusion changes and RBP activity; implemented in Python using Stan.