SVmine

SVmine refines and integrates structural variation (SV) predictions from multiple algorithms using local realignment and likelihood-based quality assessment to improve detection sensitivity and breakpoint resolution from high-throughput sequencing data.

Key Features:

• Integration of SV predictions: Mines and aggregates SV calls from various existing SV detection algorithms to create a combined candidate set.
• Local realignment refinement: Refines SV predictions by locally realigning sequencing reads around candidate SV sites.
• Likelihood-based quality assessment: Assesses SV call quality using likelihood metrics derived from the realignments.
• Sequence-context construction: Constructs sequence contexts near candidate SVs by integrating nearby single nucleotide variations, insertions, and deletions into the reference sequence.
• Sandwich alignment algorithm: Employs a sandwich alignment algorithm to further refine and improve breakpoint position estimates.
• High-throughput sequencing input: Operates on data derived from high-throughput sequencing reads.
• Improved performance on benchmarks: Demonstrates increased sensitivity, specificity, and precision of breakpoint estimations on simulated and real datasets.
• Enhanced inter-algorithm concordance: Increases overlap between SV predictions from different algorithms to improve collective detection performance.

Scientific Applications:

• Structural variation detection and characterization: Improves identification and breakpoint resolution of SVs in genomic sequences.
• Human genomics and disease studies: Supports analyses of SVs that affect genetic diversity and disease susceptibility in human genomes.
• Multi-caller consensus analysis: Enhances the combined performance of multiple SV callers by increasing prediction concordance.
• Benchmarking and validation: Provides refined SV calls useful for evaluation on simulated and real datasets.

Methodology:

SVmine mines SV predictions from existing tools, constructs local reference-derived sequence contexts by integrating nearby SNVs, insertions, and deletions, performs local realignment of sequencing reads to those contexts, computes likelihood metrics from the alignments for quality assessment, and applies a sandwich alignment algorithm to refine breakpoint positions.