BreaKmer: detection of structural variation in targeted massively parallel sequencing data using kmers.

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• Genomic structural variations (SV) are crucial in cancer but challenging to detect with targeted sequencing.
• Accurate SV detection is vital for clinical applications, including assessing actionable mutations in tumor biopsies.
• Current methods struggle with the precision required for targeted resequencing data in cancer diagnostics.

Purpose of the Study:

• To introduce BreaKmer, a novel computational approach for precise detection of genomic structural variations (SV).
• To evaluate BreaKmer's performance in identifying diverse SV types (insertions, deletions, inversions, duplications, translocations) in targeted sequencing data.
• To compare BreaKmer's sensitivity and specificity against existing algorithms for SV detection in tumor samples.

Main Methods:

• Developed BreaKmer, a kmer-based strategy to assemble misaligned sequence reads.
• Predicted SV by realigning an assembled consensus sequence from abnormally aligned reads to the reference genome.
• Validated BreaKmer using targeted resequencing data from tumor specimens, non-tumor samples, and whole-genome sequencing data.

Main Results:

• BreaKmer achieved 97.4% overall sensitivity for detecting known structural variations (SV).
• Identified 17 novel, positively validated SVs, demonstrating its capability to discover new variants.
• Outperformed four public algorithms in SV detection sensitivity and showed reduced false positive calls in non-tumor samples.

Conclusions:

• BreaKmer offers a highly sensitive and accurate method for base-pair resolution SV detection in targeted resequencing data.
• Its performance characteristics make it suitable for clinical and research settings analyzing tumor specimens.
• The approach enhances the utility of targeted sequencing for comprehensive genomic profiling in cancer.