GRIDSS: sensitive and specific genomic rearrangement detection using positional de Bruijn graph assembly

Abstract

The identification of genomic rearrangements, particularly in cancers, with high sensitivity and specificity using massively parallel sequencing remains a major challenge. Here, we describe the Genome Rearrangement IDentification Software Suite (GRIDSS), a high-speed structural variant (SV) caller that performs efficient genome-wide break-end assembly prior to variant calling using a novel positional de Bruijn graph assembler. By combining assembly, split read and read pair evidence using a probabilistic scoring, GRIDSS achieves high sensitivity and specificity on simulated, cell line and patient tumour data, recently winning SV sub-challenge #5 of the ICGC-TCGA DREAM Somatic Mutation Calling Challenge. On human cell line data, GRIDSS halves the false discovery rate compared to other recent methods. GRIDSS identifies non-template sequence insertions, micro-homologies and large imperfect homologies, and supports multi-sample analysis. GRIDSS is freely available at https://github.com/PapenfussLab/gridss.

Optical mapping reveals a higher level of genomic architecture of chained fusions in cancer

“…Appreciating the different information content between sequencing and optical mapping, we sought to reanalyze the short-read sequencing data using a new approach, GRIDSS, specifically designed to identify CGRs (Cameron et al 2017). To evaluate the optical mapping fusions, we pulled out a set of 107 unique breakpoints from the fusion maps (Supplemental Table S2).…”

“…This set of breakpoints was compared against a breakpoint call set derived from the Illumina short-read sequencing data (Supplemental Table S2). Breakpoints were identified in Illumina sequencing data by running GRIDSS version 1.3.4 using default parameters (Cameron et al 2017) on the aligned BAM files used by Garsed et al (2014). As the BAMs were aligned to GRCh37, liftOver of optical mapping coordinates from GRCh38 to GRCh37 was performed.…”

“…Specifically, sequencing has base pair resolution allowing the detection of smaller variants, whereas mapping has resolution of about 1 kb but can detect variants that are kilo-to megabases in size. To integrate the two data sets and validate optical mapping calls, we reanalyzed the short-read sequencing data using GRIDSS, a new positional de Bruijn graph-based SV calling approach (Cameron et al 2017) and traversed across chains following multiple breakpoints (for details, see Methods). We found support for 86 of 107 unique optical mapping fusions (Supplemental Table S2).…”

GRIDSS, PURPLE, LINX: Unscrambling the tumor genome via integrated analysis of structural variation and copy number

“…The first component of the toolkit is version 2 of the GRIDSS structural variant caller [15]. Like the first version, GRIDSS performs genome-wide break-end assembly followed by probabilistic variant calling based on the totality of the direct read/read pair support, and the assembly contig support.…”

“…Here, we present a comprehensive toolkit for integrated somatic copy number and structural variation analysis using short read whole genome sequencing data from tumor/normal sample pairs. Our toolkit consists of 3 highly integrated tools: GRIDSS version 2, an enhancement of the GRIDSS [15] SV caller that also performs SV chained assembly and reports single breakend variants; PURPLE, which performs SV-aware copy number segmentation and purity and ploidy estimation, and LINX, which clusters the genomic rearrangements into higher order events and interprets their impact.…”