SV-JIM

SV-JIM streamlines structural variant calling between pairs of genomes by evaluating multiple SV callers and aggregating results based on caller support using the Jaccard index measure.

Key Features:

• Multiple Caller Evaluation: SV-JIM evaluates multiple SV sets produced by different SV callers using the Jaccard index measure to identify those with the highest consistency.
• Support-Based Aggregation: SV-JIM produces aggregated SV results based on how many callers supported each reported structural variant.
• Workflow Implementation: The pipeline is implemented using the Snakemake workflow management system.
• Retention of Rare SV Types: Aggregating SV sets with minimum support requirements helps retain less frequently occurring SV types rather than relying on specific caller combinations.

Scientific Applications:

• Pairwise Genome Comparison: SV-JIM detects variations between pairs of genomes using both genome assemblies and long-reads.
• Inter-Caller Variance Assessment: SV-JIM identifies significant inter-caller variance in SV calling results, which can vary by tens of thousands of results on larger genomes.
• Precision Reporting Evaluation: SV-JIM helps identify potential inflated precision reporting that can occur during SV evaluation.

Methodology:

SV-JIM is implemented as a Snakemake workflow that takes genome assemblies and long-reads as input, runs multiple SV callers, evaluates the resulting SV sets using the Jaccard index measure, and aggregates SVs based on caller support.