MetaFusion

MetaFusion integrates outputs from multiple fusion callers to detect and prioritize RNA-seq gene fusion candidates with high confidence for cancer genomics applications.

Key Features:

• Ensemble calling: Integrates outputs from multiple fusion callers into a unified ensemble analysis.
• Common Fusion Format (CFF): Standardizes individual caller results by conversion into the Common Fusion Format (CFF) for cross-tool comparison.
• FusionAnnotator annotation: Annotates fusion calls using FusionAnnotator and cancer fusion databases to add biological context.
• Graph clustering: Uses graph clustering to merge annotations and consolidate overlapping calls.
• Filtering and prioritization: Applies rigorous filtering to prioritize high-confidence fusion candidates.
• Ranking by metrics: Ranks candidates based on precision and recall metrics and reports performance on real and simulated datasets, including up to 100% precision in select cases.
• Benchmarking toolkit: Includes a benchmarking toolkit to calibrate and evaluate new fusion callers.

Scientific Applications:

• Fusion detection and prioritization: Detection and prioritization of RNA-seq gene fusion candidates in cancer genomics studies.
• Caller benchmarking: Calibration and comparative evaluation of fusion-calling algorithms using the included benchmarking toolkit.
• Annotation-informed interpretation: Integration of cancer fusion database annotations via FusionAnnotator to support biological relevance assessment.

Methodology:

Converts individual caller outputs to the Common Fusion Format (CFF), annotates calls with FusionAnnotator and cancer fusion databases, merges calls by graph clustering, applies filtering and ranks candidates by precision and recall, and implements ensemble calling alongside a benchmarking toolkit for caller calibration.