GASOLINE

GASOLINE detects germline and somatic structural variants from long-read sequencing data for high-resolution identification of complex genomic structural variation.

Key Features:

• Comprehensive Variant Detection: Detects germline SVs from single samples and somatic SVs from paired test-control samples.
• Long-read Sequencing Support: Operates on long-read sequencing data spanning tens to hundreds of kilobases for improved resolution of complex SVs.
• Sophisticated Clustering Procedure: Groups SV signatures using a modified reciprocal overlap criterion to improve clustering beyond read-depth and intra- and inter-alignment signature approaches.
• Multi-language Implementation: Implemented using Perl, R, and Fortran.
• Efficient Data Handling: Processes aligned data in BAM format and outputs VCF files containing statistically significant somatic SVs.
• Performance: Reports run times of approximately 4–5 hours for 30× sequencing coverage using 20 threads.

Scientific Applications:

• WGS long-read method benchmarking: Serves as a comparative method for whole-genome sequencing long-read computational SV detection workflows.
• Cancer somatic SV discovery: Enables detection of somatic SVs in tumor-normal pairs, demonstrated by identification of five somatic SVs in metastatic melanoma matched-normal samples that were missed by five other sequencing technologies.

Methodology:

Processes aligned BAM files, groups SV signatures using a modified reciprocal overlap clustering procedure, analyzes single samples for germline and paired samples for somatic SVs, and outputs VCF files with statistically significant somatic variants.