SV-Bay

SV-Bay detects structural variants in whole-genome sequencing data from mate-pair or paired-end reads using a probabilistic Bayesian framework to improve sensitivity and specificity for identifying large somatic rearrangements in cancer genomes.

Key Features:

• Integration of Normal and Abnormal Reads: Combines depth of coverage by normal reads with abnormally mapped read pairs to characterize large somatic rearrangements.
• Probabilistic Bayesian Framework: Models the likelihood of structural variants with a Bayesian probabilistic approach that integrates multiple genomic signals.
• Correction for GC-content and Read Mappability: Applies corrections for GC-content and read mappability to adjust expected read counts and mitigate biases in variant detection.
• Somatic Variant Detection with Matched Normal Samples: Utilizes matched normal samples when available to distinguish somatic from germline variants.

Scientific Applications:

• Characterization of Somatic Rearrangements: Identifies and refines calls for large somatic rearrangements in cancer genomes using mate-pair and paired-end data.
• Cancer Genomics Studies: Supports detection and interpretation of somatic structural variants for tumor genomics research and comparison across samples.
• Benchmarking and Method Comparison: Enables method evaluation and benchmarking using simulated datasets and experimental mate-pair data.

Methodology:

Uses a Bayesian probabilistic model that integrates depth of coverage and abnormally mapped read pairs, applies GC-content and read mappability corrections, optionally incorporates matched normal samples, and validates predictions against simulated datasets and experimental mate-pair data from the CLB-GA neuroblastoma cell line, demonstrating improved sensitivity and a lower false-positive rate.