SVFX

SVFX assigns pathogenicity scores to structural variants (SVs), including large deletions and duplications, using a mechanism-agnostic machine-learning approach to prioritize SVs by predicted pathogenic impact.

Key Features:

• Machine learning framework: Uses Random Forest algorithms to assign pathogenicity scores and prioritize SVs.
• Mechanism-agnostic workflow: Applies a mechanism-agnostic workflow to score SVs independent of assumed molecular mechanisms.
• SV types: Targets structural variants including large deletions and duplications.
• Somatic and germline models: Generates predictive models for both somatic and germline structural variants.
• Feature integration: Incorporates genomic, epigenomic, and conservation-based attributes derived from SV call sets in diseased and healthy individuals.
• Pathogenic vs benign discrimination: Trains on comprehensive datasets to distinguish pathogenic SVs from benign variants.
• Cancer enrichment: When applied to cancer cohorts, enriches for known cancer genes and pathways associated with cancer-related SVs.

Scientific Applications:

• SV prioritization in disease studies: Prioritizes structural variants by predicted pathogenic impact for disease-relevant analyses.
• Somatic and germline SV analysis: Enables separate analysis and modeling of somatic and germline SVs in cohort studies of diseased and healthy individuals.
• Cancer genomics: Identifies and enriches for cancer genes and pathways associated with structural variants in cancer cohorts.

Methodology:

Employs a mechanism-agnostic workflow using Random Forest algorithms trained on genomic, epigenomic, and conservation-based features derived from SV call sets in diseased and healthy individuals.