Somatic Structural Variant Discovery from Tumor Specific Strings Computed Against a Draft Composite Germline Reference

Somatic structural variants, crucial in cancer, are tough to detect accurately due to complex genomic rearrangements, repetitive regions, and sequencing errors. Current methods struggle with alignment issues in repetitive areas and near SV breakpoints, or risk passing on errors from multiple datasets. Researchers have devised VESPA-SS, a fresh method to identify somatic SVs from tumor-normal long-read whole-genome sequencing data.

This approach leverages specific strings derived from tumor reads against a Draft Composite Germline Reference, avoiding initial mapping or top-notch germline assembly. After extracting signals, VESPA-SS clusters and refines them using consensus sequences and alignment-based techniques to pinpoint somatic SVs. Tests on real PacBio HiFi data show VESPA-SS outperforms existing callers, scoring higher on F1 scores for HG008 and COLO829 cancer cell lines.

This breakthrough, detailed in the study, paves the way for a new computational method to detect complex genomic changes in cancer, available publicly for use.