annoFuse

annoFuse annotates and prioritizes putative oncogenic RNA fusions from RNA-Seq fusion caller outputs to identify biologically and clinically relevant fusion events.

Key Features:

• Standardized Filtering: Applies filtering criteria including requiring at least one junction read and removing fusions where spanning fragment support exceeds ten times the junction read count to reduce artifactual calls.
• Prioritization of Fusions: Prioritizes fusions present in TCGA or involving gene partners annotated as oncogenes, tumor suppressor genes, COSMIC genes, or transcription factors.
• Annotation Capabilities: Uses FusionAnnotator annotations to filter non-oncogenic and artifactual fusions, annotates protein domains using PFAM, assesses fusion reciprocality, and identifies kinase domain retention in gene partners.
• Recurrent Fusion Analysis: Determines recurrent fusions across cohorts and identifies genes that are frequently fused within specific histologies.
• Integration with Multiple Callers: Integrates and merges fusion calls from multiple RNA-Seq fusion callers including STAR-Fusion and Arriba.

Scientific Applications:

• TCGA validation: Validated on a random cohort of TCGA RNA-Seq samples with 96% sensitivity in retaining high-confidence fusions.
• OpenPBTA pediatric brain tumor analysis: Applied to OpenPBTA pediatric brain tumor RNA-Seq samples to identify recurrent fusions and fused genes across different brain tumor histologies.

Methodology:

Filters fusion calls using FusionAnnotator annotations and read-support criteria (requiring at least one junction read and removing calls where spanning fragment support >10× junction reads), merges calls from STAR-Fusion and Arriba, annotates protein domains with PFAM, assesses fusion reciprocality and kinase domain retention, prioritizes based on TCGA presence and involvement of oncogenes, tumor suppressor genes, COSMIC genes, or transcription factors, and determines recurrence across cohorts.