James Han

Detecting low variant allele fraction (VAF) mosaic variants without matching controls remains a major challenge in genomics, limited by technical noise, lack of benchmarks, and computational scalability. We present the DRAGEN mosaic caller, a hardware-accelerated approach identifying variants down to ∼1-2% VAF with low false-positive rates and hour-scale runtimes for mosaic SNV/indel detection from bulk sequencing. To support evaluation, we introduce a genome-wide low-VAF benchmark for variants between 1-10% VAF. Application to blood, sperm, and brain tissues revealed patterns, including mosaic hotspots and mutational signatures. The first analysis of HG002 blood showed that many “mosaic” variants defined from HG002 cell lines are likely culture-derived and not in vivo mutations. Importantly, DRAGEN also enables personalized assembly pangenome references to improve alignment and mosaic variant detection in complex regions. Together, this development makes routine low-VAF discovery feasible, opening new opportunities to study mosaic mutations in healthy and disease individuals.

Accurate and comprehensive genetic characterization of acute myeloid leukemia (AML) is essential for diagnosis, prognostication, and treatment selection. We report here, in 255 adults with AML enrolled in a prospective clinical protocol at 18 major cancer centers across the USA, the results of whole genome DNA-sequencing (WGS) at diagnosis and post-treatment remission. WGS effectively recapitulated, and frequently identified genetic alterations missed by, conventional standard of care clinical testing. These new findings included important prognostic and predictive biomarkers, copy number alterations, regulatory element, splicing, and structural variants including partial tandem duplications within KMT2A. All patients had a pathogenic variant detected at diagnosis, and approximately ten percent also had evidence of a potential inherited myeloid malignancy predisposition. This comprehensive atlas of adult AML genomics provides novel insights into disease biology, creates an evidentiary basis to support clinical testing improvements, and is a resource for both diagnostics and drug development. . Statement of Significance: Acute myeloid leukemia is a diagnostic category encompassing multiple rare hematological malignances. We show, in this nationwide multicenter study, that standardized unbiased whole genome DNA-sequencing and disease-optimized bioinformatics can replicate conventional “standard of care” AML clinical testing results, while also revealing currently underdiagnosed AML disease biology and potential genetic predisposition.

Despite significant advances in genomic sequencing, the resolution of many rare disease cases is still hindered by variant detection limitations. Short reads struggle in homologous regions, and long reads remain costly and difficult to scale. Here, we present the first systematic evaluation of Illumina’s Constellation sequencing technology for rare disease research. By fragmenting long DNA molecules directly on the flow cell surface, Constellation unlocks proximity information that enables long-range phasing and structural variant detection. Across 21 families, Constellation independently identified all known causative variants and resolved previously unsolved trios. It reliably resolved complex structural and copy number variants (e.g. impacting MECP2 ) and recovered haplotype phasing information across key disease impacting variants, all from low DNA input using existing Illumina infrastructure. These findings establish Constellation as a scalable, cost-efficient advance, closing critical diagnostic gaps and broadening access to long-range variant analysis in rare disease genomics.

Motivation: Whole-genome sequencing (WGS) is increasingly preferred for clinical applications due to its comprehensive coverage, effectiveness in detecting copy number variants (CNVs), and declining costs. However, systematic evaluations of WGS CNV callers tailored to germline clinical testing—where high sensitivity and confirmation of reported CNVs are essential—remain necessary. Clinical reporting typically emphasizes CNVs affecting coding regions over precise breakpoint detection. This study benchmarks several short-read WGS CNV detection tools using reference cell lines to inform their clinical use. Results: While tools vary in sensitivity (7%–83%) and precision (1%–76%), few meet the sensitivity needed for clinical testing. Callers generally perform better for deletions (up to 88% sensitivity) than duplications (up to 47% sensitivity), with poor detection of duplications under 5 kb. Notably, for CNVs in genes commonly included in clinical panels, significantly improved sensitivity and precision were observed when benchmarking against 25 cell lines with known CNVs. DRAGEN v4.2 high-sensitivity CNV calls, post-processed with custom filters, achieved 100% sensitivity and 77% precision on the optimized gene panel after excluding recurring artifacts. This level of performance may support clinical use with orthogonal confirmation of reportable CNVs, pending validation on laboratory-specific samples. Availability and implementation: The data underlying this article are available in the European Nucleo-tide Archive under project accession PRJEB87628.

Somatic mosaicism is an important cause of disease, but mosaic and somatic variants are often challenging to detect because they exist in only a fraction of cells. To address the need for benchmarking subclonal variants in normal cell populations, we developed a benchmark containing mosaic variants in the Genome in a Bottle Consortium (GIAB) HG002 reference material DNA from a large batch of a normal lymphoblastoid cell line. First, we used a somatic variant caller with high coverage (300x) Illumina whole genome sequencing data from the Ashkenazi Jewish trio to detect variants in HG002 not detected in at least 5% of cells from the combined parental data. These candidate mosaic variants were subsequently evaluated using >100x BGI, Element, and PacBio HiFi data. High confidence candidate SNVs with variant allele fractions above 5% were included in the HG002 draft mosaic variant benchmark, with 13/85 occurring in medically relevant gene regions. We also delineated a 2.45 Gbp subset of the previously defined germline autosomal benchmark regions for HG002 in which no additional mosaic variants >2% exist, enabling robust assessment of false positives. The variant allele fraction of some mosaic variants is different between batches of cells, so using data from the homogeneous batch of reference material DNA is critical for benchmarking these variants. External validation of this mosaic benchmark showed it can be used to reliably identify both false negatives and false positives for a variety of technologies and detection algorithms, demonstrating its utility for optimization and validation. By adding our characterization of mosaic variants in this widely-used cell line, we support extensive benchmarking efforts using it in simulation, spike-in, and mixture studies.

Research and medical genomics require comprehensive, scalable methods for the discovery of novel disease targets, evolutionary drivers and genetic markers with clinical significance. This necessitates a framework to identify all types of variants independent of their size or location. Here we present DRAGEN, which uses multigenome mapping with pangenome references, hardware acceleration and machine learning-based variant detection to provide insights into individual genomes, with ~30 min of computation time from raw reads to variant detection. DRAGEN outperforms current state-of-the-art methods in speed and accuracy across all variant types (single-nucleotide variations, insertions or deletions, short tandem repeats, structural variations and copy number variations) and incorporates specialized methods for analysis of medically relevant genes. We demonstrate the performance of DRAGEN across 3,202 whole-genome sequencing datasets by generating fully genotyped multisample variant call format files and demonstrate its scalability, accuracy and innovation to further advance the integration of comprehensive genomics. Overall, DRAGEN marks a major milestone in sequencing data analysis and will provide insights across various diseases, including Mendelian and rare diseases, with a highly comprehensive and scalable platform.

Whole-genome sequencing (WGS) is increasingly favored over other genomic sequencing methods for clinical applications due to its comprehensive coverage and declining costs. WGS is particularly useful for the detection of copy number variants (CNVs), presumed to be more accurate than targeted sequencing assays such as WES or gene panels, because it can identify breakpoints in addition to changes in coverage depth. Recent advancements in bioinformatics tools, including those employing hardware acceleration and machine learning, have enhanced CNV detection. Although numerous benchmarking studies have been published, primarily focusing on open-source tools for short-read WGS CNV calling, systematic evaluations that encompass commercially available tools that meet the rigorous demands of clinical testing are still necessary. In clinical settings, where the confirmation of reported CNVs is often required, there is a higher priority on sensitivity over specificity/precision compared to research applications. Moreover, clinical gene panel reporting primarily concerns whether a CNV affects coding regions or, in some cases, promoters, rather than the precise detection of breakpoints. This study aims to benchmark the performance of various CNV detection tools tailored for clinical reporting from WGS using reference cell lines, providing insights critical for optimizing clinical diagnostics. Our results indicate that while different tools exhibit strengths in either sensitivity or precision and are better suited for certain classes and lengths of variants, few can deliver the balanced performance essential for clinical testing, where high sensitivity is imperative. Generally, callers demonstrate better performance for deletions than duplications, with the latter being poorly detected in events shorter than 5kb. We demonstrate that the DRAGEN™ v4.2 CNV caller, particularly with custom filters on its high sensitivity mode, offers a superior balance of sensitivity and precision compared to other available tools.

Research and medical genomics require comprehensive and scalable solutions to drive the discovery of novel disease targets, evolutionary drivers, and genetic markers with clinical significance. This necessitates a framework to identify all types of variants independent of their size (e.g., SNV/SV) or location (e.g., repeats). Here we present DRAGEN that utilizes novel methods based on multigenomes, hardware acceleration, and machine learning based variant detection to provide novel insights into individual genomes with ∼30min computation time (from raw reads to variant detection). DRAGEN outperforms all other state-of-the-art methods in speed and accuracy across all variant types (SNV, indel, STR, SV, CNV) and further incorporates specialized methods to obtain key insights in medically relevant genes (e.g., HLA, SMN, GBA). We showcase DRAGEN across 3,202 genomes and demonstrate its scalability, accuracy, and innovations to further advance the integration of comprehensive genomics for research and medical applications.

Rhizobium leguminosarum bv. trifolii strain TA1 is an aerobic, motile, Gram-negative, non-spore-forming rod that is an effective nitrogen fixing microsymbiont on the perennial clovers originating from Europe and the Mediterranean basin. TA1 however is ineffective with many annual and perennial clovers originating from Africa and America. Here we describe the features of R. leguminosarum bv. trifolii strain TA1, together with genome sequence information and annotation. The 8,618,824 bp high-quality-draft genome is arranged in a 6 scaffold of 32 contigs, contains 8,493 protein-coding genes and 83 RNA-only encoding genes, and is one of 20 rhizobial genomes sequenced as part of the DOE Joint Genome Institute 2010 Community Sequencing Program.