Journal: medRxiv

Article Title: Long-Read Haplotype Phasing Resolves Allelic Configuration as a Missing Layer of Precision Oncology

doi: 10.64898/2026.05.05.26351600

Figure Lengend Snippet: (A) Clinical samples undergo standard short-read whole-exome and targeted panel sequencing a part of routine precision oncology care. Cases are nominated for long-read nanopore sequencing based on three complementary screening criteria: RNA evidence (outlier gene expression, rare splice junctions, viral transcript detection); DNA evidence (cryptic driver candidates, rare mutational signatures, complex copy number architecture); and relevant medical history (family history of cancer, early disease onset, unusual clinical course). Long-read bioinformatics analysis simultaneously characterizes complex structural rearrangements, performs haplotype phasing, and detects DNA methylation directly from base modification signals. Findings are communicated as amended clinical reports to treating oncologists. (B) The timeline illustrates the complete workflow from sample receipt to amended report, with a median turnaround of 8–20 days.

Article Snippet: Sequencing libraries were prepared using the Ligation Sequencing Kit V14 (SQK-LSK114; Oxford Nanopore Technologies) in combination with the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing (E7180S; New England Biolabs), following the manufacturers’ protocols.

Techniques: Sequencing, Nanopore Sequencing, Gene Expression, DNA Methylation Assay, Modification

Journal: medRxiv

Article Title: Long-Read Haplotype Phasing Resolves Allelic Configuration as a Missing Layer of Precision Oncology

doi: 10.64898/2026.05.05.26351600

Figure Lengend Snippet: (A) Molecular diagnostic workflow for the PROBLEM cohort, showing computational mutational signature-guided nomination of 164 cases with genomic instability phenotypes from 768 patients with prostate, breast, or ovarian cancers. (B) Representative cryptic MMRD mechanism: allele-specific methylation tracks across the MLH1 promoter demonstrating coordinated biallelic hypermethylation in two cases (MO_1444, MO_1533) relative to an unmethylated control (MO_1789). (C) Long-read structural variant characterization of a chromothripsis-like PMS2 rearrangement in a neuroendocrine prostate cancer case (MO_2707). Short-read copy number analysis identified a hemizygous deletion of the PMS2 region but could not resolve the second hit, which was characterized only by long-read sequencing. (D) Representative cryptic FTD mechanism: haplotype phasing enables detection of biallelic CDK12 inactivation in MO_2674. One allele harbors a 16 kb deletion removing the last exon and 3’ UTR, likely contributing to nonsense-mediated decay. The opposing allele contains an inversion truncating the first 20 kb of CDK12 and a duplication of the orphaned segment. Haplotype-specific methylation analysis demonstrates progressive loss of epigenetic maintenance across the orphaned copies, with methylation levels declining toward the 3’ region. (E) Top : Patient MO_1159 harbors NM_024675.4:c.2835-12T>G, an ultra-rare novel splice PALB2 intronic variant. SpliceAI computational prediction demonstrates 94% probability that this germline variant abolishes the canonical splice acceptor site through disruption of the polypyrimidine tract. Haplotype phasing demonstrated that the germline variant occurs in trans configuration with a somatic nonsense mutation (p.Gln39Ter), confirming biallelic PALB2 inactivation. Bottom : Validated RNA sequencing analysis demonstrated that this mutation triggers an exon skipping event.

Article Snippet: Sequencing libraries were prepared using the Ligation Sequencing Kit V14 (SQK-LSK114; Oxford Nanopore Technologies) in combination with the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing (E7180S; New England Biolabs), following the manufacturers’ protocols.

Techniques: Diagnostic Assay, Methylation, Control, Variant Assay, Sequencing, Disruption, Mutagenesis, RNA Sequencing

Journal: medRxiv

Article Title: Long-Read Haplotype Phasing Resolves Allelic Configuration as a Missing Layer of Precision Oncology

doi: 10.64898/2026.05.05.26351600

Figure Lengend Snippet: (A) Alluvial diagram illustrating the flow of diagnostic resolution across MMRD, FTD, and HRD categories in prostate, breast, and ovarian cancers, from initial signature classification through short-read and long-read sequencing, including reclassification of 11 ambiguous HRD cases as HRD-like aetiologies following secondary signature analysis. (B) Heatmap displays structural variant features including insertions and deletions stratified by length, extracted using the SigProfilerExtractor package. SigProfilerExtractor categorizes structural variants as clustered or non-clustered; for this analysis, only non-clustered events were utilized to prevent artificial clustering bias from chromothripsis-associated “clustered” samples. Features underwent log2 transformation to normalize the wide range of structural variant counts and enable comparison across different event types. Manhattan distance calculation was employed because it is more robust to outliers and better suited for count-based data, particularly when analyzing discrete structural variant events. Complete linkage clustering methodology was applied to maximize between-cluster separation and create distinct, well-separated mutational pattern groups. Additional annotation tracks provide complementary mutational signature information: COSMIC indel signatures ID6 and ID8 (classified as positive when contribution ≥20%), quantification of microhomology sequences ≥3bp flanking structural variant breakpoints, and HRDetect scores for homologous recombination deficiency assessment. The analysis incorporates positive control cases with known driver alterations (indicated as red box) identified through short-read sequencing and negative control cases (indicated as gray box in annotation track) definitively lacking genomic instability signatures, providing validation benchmarks to guide interpretation of cryptic mutational mechanisms. Cases with pathogenic germline variants but lacking HRD phenotype were marked with red asterisks. For these analyses, tumor samples without matched normal controls were excluded. All FTD cases clustered together due to their characteristic tandem duplications ranging from 0.1 MB to 10 MB. Overall, breast cancer and ovarian cancer cases with HRD caused by BRCA1 / BARD1 alterations also clustered together, reflecting their enrichment for short tandem duplications and short deletions. Some HRD samples with BRCA2 and PALB2 deficiency showed less genomic scarring, characterized primarily by enrichment for short deletions. MMRD cases lacked characteristic copy number signatures, consistent with their primary manifestation through point mutations rather than structural alterations. (C) Oncoprint of gene-level mechanisms resolved across HRD, FTD, and MMRD cases, distinguishing events resolved by short-read sequencing (gray), resolved by long-read sequencing (teal), and providing additional mechanistic insight through long-read analysis (blue). Asterisks denote pathogenic germline variants.

Article Snippet: Sequencing libraries were prepared using the Ligation Sequencing Kit V14 (SQK-LSK114; Oxford Nanopore Technologies) in combination with the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing (E7180S; New England Biolabs), following the manufacturers’ protocols.

Techniques: Diagnostic Assay, Sequencing, Variant Assay, Transformation Assay, Comparison, Homologous Recombination, Positive Control, Negative Control, Biomarker Discovery

Journal: medRxiv

Article Title: Long-Read Haplotype Phasing Resolves Allelic Configuration as a Missing Layer of Precision Oncology

doi: 10.64898/2026.05.05.26351600

Figure Lengend Snippet: (A) Frequency of multi-hit events across curated cancer genes in 4,496 tumor samples (3,688 patients). Bars represent the percentage of samples with ≥2 somatic mutations among all samples harboring any mutation in that gene, ordered by decreasing frequency. Blue, tumor suppressor; red, oncogene; amber, context-dependent. (B) VAF correlation of co-occurring SNV pairs within the same gene and sample (n = 12,934). Hexagonal bins colored by log-scaled pair count; white bins indicate zero observations. Dashed line, VAF□= VAF□. Pearson r reported. (C) Genomic distance distribution of co-occurring SNV pairs (n = 12,934; log-scaled x-axis). Blue, phasable by short-read sequencing (≤500 bp); red, requiring long-read sequencing (>500 bp). (D) NOTCH1 protein schematic showing compound cis variant pairs (teal bars) confirmed by long-read phasing across six patients, consistently pairing heterodimerization (HD) domain with PEST domain variants. (E) Luciferase reporter assays measuring transcriptional activation of compound NOTCH1 HD+PEST mutations relative to individual variants. Four of six compound pairs showed significantly enhanced NOTCH pathway activation. Unpaired two-tailed Welch’s t-test; n = 3 per group; *p < 0.05, **p < 0.01, ***p < 0.001. (F) PIK3CA protein schematic (ABD, RBD, C2, helical, kinase domains) showing compound cis variant pairs confirmed by long-read phasing in breast cancer and additional tumor types. Each row represents one patient; circles denote variant positions; dashed lines indicate confirmed cis configuration. (G) PDGFRB protein schematic showing compound cis variants in three mesenchymal tumors: two infantile myofibromatosis cases with germline(†) p.Arg561Cys paired with somatic Asn666 alterations, and one sporadic myofibroma with dual somatic tyrosine substitutions. (H) KIT protein schematic showing compound cis variants across three cases, spanning juxtamembrane and activation loop domains.

Article Snippet: Sequencing libraries were prepared using the Ligation Sequencing Kit V14 (SQK-LSK114; Oxford Nanopore Technologies) in combination with the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing (E7180S; New England Biolabs), following the manufacturers’ protocols.

Techniques: Mutagenesis, Sequencing, Variant Assay, Luciferase, Activation Assay, Two Tailed Test