Characterization of multiple myeloma genomes with LinkPrep™ assay enables sensitive detection of somatic variation and SV-driven interactions of the 3D genome. Free

Introduction:

Multiple myeloma (MM) is characterized by its genomic complexity that includes recurrent driver single nucleotide variation (SNVs) alongside a high burden of structural variation (SV). Complete characterization of somatic variation in a MM genome is dependent on a technology that can accurately and sensitively detect both small and large variants. While traditional short read sequencing is the gold standard for characterization of simple SNVs, it has limited accuracy for characterization of SVs especially those involving complex rearrangements and challenging to map loci (i.e. IGH). Here, we introduce LinkPrep™ technology, a novel linked-read approach compatible with short read sequencers that delivers high accuracy for both small and large somatic variant detection. We demonstrate LinkPrep performance across four separate relapse MM genomes.

Methods:

Four MM patient-derived xenografts were profiled with the Dovetail® LinkPrep™ kit and sequenced to ~30X on an Illumina platform. Data was analyzed through the Dovetail Analysis Portal to detect SNVs/Indels (DeepSomatic), CNVs (Purple), and SVs (HiC-breakfinder and proprietary tools). Call sets were compared against other genomics technologies including WGS (80X tumor/ 30X normal), PacBio (15X), and OGM (400X). We focus analysis on one sample (M24) containing complex SV events and several driver SNVs. Using this sample, we highlight the capabilities of LinkPrep data to detect SNV/Indels, CNVs, and SVs and to link this information together to provide a candidate solution for the complex templated insertions. We further describe SV-driven 3D chromatin alterations (i.e. neo-loops, potential enhancer hijacking) from the same 30X LinkPrep dataset.

Results:

LinkPrep + Dovetail Analysis Portal (~8 hours analysis time) returned high quality variant results for the four MM samples. LinkPrep data recalled all pathogenic SNV/Indels detected by 80X WGS (notable pathogenic variants were detected in NRAS, KRAS, TP53). LinkPrep data further detected SVs involving the IGH locus (inversions and translocations) in all four samples. Deeper analysis of one sample (M24) highlights that LinkPrep data detects genetic alterations common in MM including SNV mutations in NRAS and TP53 (both occurring at 100% VAF), IGH rearrangements, chr 1q gain, chr 1p loss of heterozygosity, trisomy of chr 5,7, and 15, and complex rearrangements involving chromosomes 2, 3, 4, 16, and 19. With the exception of a confident subclonal (1% SV-VAF) unbalanced translocation detected by LinkPrep, the remaining LinkPrep large SV calls were validated by other genomics methods (tumor/normal 80X WGS/15X PacBio/400X OGM). Underscoring the capability of LinkPrep for SV detection, LinkPrep calls at 30X contain >10 times more read support for every SV call over other technologies. Using the linked-read feature of LinkPrep data along with CNV segment calls and SV calls provided through the Dovetail Analysis Portal a candidate solution is derived for the complex rearrangement involving chromosomes 2, 3, 4, 16, and 19. This LinkPrep solution, refined down to base pair resolution for every breakpoint, is verified by 400X OGM data. Uniquely, LinkPrep data further enables visualization of the SV-reconstructed 3D genome and reveals neo-loops connecting enhancers with promoters to suggest enhancer hijacking mechanisms associated with the complex rearrangement.

Conclusion:

We applied and validated LinkPrep technology, a novel, multi-dimensional, linked-read method, to a MM sample cohort. The technology enables high accuracy comprehensive variant detection, offering improved sensitivity and accuracy for SV detection over short and long reads, including those in challenging genomic regions such as IGH. Linked-reads captured in LinkPrep data enable phased reconstructions of complex rearrangements often found in MM genomes. Furthermore, the same LinkPrep data enables detection of epigenetic state and neo-3D genome interactions driven by SVs to elucidate enhancer hijacking events. Together, these data demonstrate that LinkPrep linked-reads enable improved characterization of somatic variation in highly rearranged genomes, while simultaneously detecting epigenetic states – all enabled with the accuracy and cost-effectiveness of short read sequencing.