OncoSplicing

OncoSplicing catalogs clinically relevant alternative splicing (AS) events from TCGA SpliceSeq and TCGA SplAdder to support analyses that link AS to clinical outcomes across human cancers.

Key Features:

• Integrated datasets: Integrates TCGA SpliceSeq (122,423 AS events in 33 cancer types) and TCGA SplAdder (238,558 AS events in 32 cancer types).
• Survival Analysis: Performs survival analysis using both median and optimal cut-offs to assess prognostic significance of AS events.
• Differential Splicing Analysis: Conducts differential analysis between TCGA tumor samples and adjacent normal or GTEx normal samples.
• Pan-Cancer Views: Provides pan-cancer summaries of AS patterns, differences, and Cox proportional hazards regression results.
• Clinical Indicator-Relevant Splicing Events: Identifies AS events associated with clinical indicators for potential diagnostic or prognostic relevance.
• Data Accessibility: Enables download of splicing data derived from the SplAdder project.
• Visualization Tools: Generates Kaplan–Meier plots of AS events across 31 cancers.
• Prognostic Prediction Signatures: Builds prognostic signatures using a random forest survival algorithm based on survival-associated AS events across 31 cancer types.
• Molecular Subtype Classification: Explores molecular subtype classification using consensus K-means clustering analysis of AS events.

Scientific Applications:

• Mechanistic studies of oncogenesis: Investigate how aberrant alternative splicing contributes to cancer development and progression.
• Biomarker discovery: Link AS events to clinical outcomes to identify candidate diagnostic and prognostic biomarkers.
• Prognostic model development and subtype analysis: Develop and evaluate prognostic signatures and molecular subtype classifications based on AS patterns.

Methodology:

Integrates AS events from TCGA SpliceSeq and TCGA SplAdder; applies survival analysis using median and optimal cut-offs, Kaplan–Meier plotting, Cox proportional hazards regression, differential splicing comparisons between TCGA tumor and adjacent or GTEx normal samples, random forest survival modeling for prognostic signatures, and consensus K-means clustering for subtype analysis; provides downloadable SplAdder-derived splicing data.