henhaobio
Sign in

Nature GeneticsGeneticsPeer-reviewed journal article

Synchronized long-read genome, methylome, epigenome, and transcriptome for resolving a Mendelian condition

Mitchell Robert VollgeriD, Jonas KorlachiD, Kiara C. EldrediD, Elliott G. SwansoniD, Jason G. Underwood, Stephanie C. BohaczukiD, Yizi MaoiD, Yong-Han Hank ChengiD, Jane E. RanchalisiD, Elizabeth BlueiD, Ulrike Schwarze, Katherine M. MunsoniD, Christopher T. SaundersiD, Aaron M. WengeriD, Aimee Allworth, Sirisak Chanprasert, Brittney L. Duerden, Ian A. GlassiD, Martha J. Horike-PyneiD, Michelle Kim, Kathleen A. LeppigiD, Ian J. McLaughliniD, Jessica T. OgawaiD, Elisabeth A. RosenthaliD, Sam Sheppeard, , Andrew B. StergachisiD

PDFDOIPreprintCode
In-depth full-text digest available · Jump to digest ↓

Abstract

Resolving the molecular basis of a Mendelian condition (MC) remains challenging owing to the diverse mechanisms by which genetic variants cause disease. To address this, we developed a synchronized long-read genome, methylome, epigenome, and transcriptome sequencing approach, which enables accurate single-nucleotide, insertion-deletion, and structural variant calling and diploid de novo genome assembly, and permits the simultaneous elucidation of haplotype-resolved CpG methylation, chromatin accessibility, and full-length transcript information in a single long-read sequencing run. Application of this approach to an Undiagnosed Diseases Network (UDN) participant with a chromosome X;13 balanced translocation of uncertain significance revealed that this translocation disrupted the functioning of four separate genes (NBEA, PDK3, MAB21L1, and RB1) previously associated with single-gene MCs. Notably, the function of each gene was disrupted via a distinct mechanism that required integration of the four omes to resolve. These included nonsense-mediated decay, fusion transcript formation, enhancer adoption, transcriptional readthrough silencing, and inappropriate X chromosome inactivation of autosomal genes. Overall, this highlights the utility of synchronized long-read multi-omic profiling for mechanistically resolving complex phenotypes.

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (27 Sept 2023) · CC BY 4.0

Pooling Fiber-seq gDNA with concatenated MAS-Seq cDNA in one PacBio run gives phased genome, methylome, chromatin, and transcriptome, exposing four gene disruptions from one X;13 translocation.

Diagnosing Mendelian conditions is limited less by variant detection than by establishing functional consequence, and existing multi-omic approaches rely on short reads and separately prepared samples, preventing haplotype-level readouts and requiring batch correction. One long-read run yielding phased sequence, CpG methylation, chromatin accessibility, and full-length transcripts lets the two haplotypes serve as internal controls for non-coding and structural variants.

Key results

  1. ShownThe pooled run still supported accurate small and structural variant calling, read phasing, and diploid de novo assembly on Genome-in-a-Bottle (GIAB) reference cells.

    GM24385 vs HG002 GIAB truth sets: SNV F1 99.94%, indel F1 98.81%, SV F1 92.5%; 90.0% of Fiber-seq reads phased; haplotype contig N50s >20 Mb at QV 50. Each sample gave over 6 million reads, 5-10% from MAS-Seq cDNA, with 26-30x genome coverage at a 15:1 molar pooling ratio. · Synchronized long-read multi-ome profiling; Genetic and epigenetic validation of…; Identification of genomic variants

  2. ShownCpG methylation, nucleosome and accessibility maps, and full-length isoforms from the same run agreed with orthogonal references, and phasing exposed haplotype-specific regulatory elements, including at imprinted loci.

    GM24385 CpG methylation Pearson correlation 0.92-0.93 vs whole-genome bisulfite sequencing (WGBS) and Oxford Nanopore; ∼332 million nucleosome footprints; 84,121 accessible elements, 97.3% in ENCODE regulatory elements; over 2 million full-length transcripts from 48,967 isoforms of 11,512 genes. GM12878: 57 elements with significant haplotype-specific accessibility. · Genetic and epigenetic validation of…; Identification of CpG methylation; Identification of chromatin architectures

  3. ShownIn fibroblasts from proband UDN318336, assembly placed the 13q breakpoint inside NBEA, so full-length NBEA transcripts arise only from the intact chromosome 13, and long-read cDNA revealed a PDK3-MAB21L1 fusion transcript on der(13).

    28x phased coverage and 2,196,793 full-length transcripts; assembled breakpoints give a ∼50 Mb der(13) and ∼210 Mb der(X); the first 41 of 58 NBEA exons lie on der(13), consistent with nonsense-mediated decay of truncated transcripts; the fusion joins PDK3 isoform 2 exon 11 to MAB21L1 isoform 2 exon 2, adding 66 amino acids. · Long-read genome exposes NBEA haploinsufficiency; Long-read transcriptome exposes PDK3-MAB21L1…; Sanger validation of der(X) fusion junction

  4. SuggestedFusion readthrough silences MAB21L1 on der(13), while a downstream retina-selective MAB21L1 enhancer escapes silencing and now sits near the PDK3 promoter, consistent with enhancer adoption raising PDK3 protein in patient retinal organoids.

    Fibroblasts: der(13)-specific MAB21L1 promoter CpG hypermethylation and transcript imbalance; retinal organoids: der(13) promoter hypermethylated with reduced accessibility. Element 9 (of nine within 10 kb) is co-accessible with the promoter, open on 76% of der(13) fibers, ∼70 kb from PDK3; PDK3 protein higher than control organoids (p 0.0183, t test). · Long-read CpG methylome exposes MAB21L1…; Long-read chromatin epigenome exposes PDK3…; Identification of MAB21L1 putative enhancer…; Figure legends

  5. SuggestedHaplotype-resolved accessibility shows the intact X is preferentially inactivated, yet X chromosome inactivation (XCI) spreads into the chromosome 13 segment of der(X) in a minority of cells, silencing RB1 as a proposed first hit for bilateral retinoblastoma.

    Fibroblasts show markedly skewed XCI favoring inactivation of chrXint; the 13p13.3-13qter segment carried on der(X) and chr13int shows significant accessibility imbalance (Mann-Whitney U test) consistent with silencing in ∼3% of cells, reaching RB1 13.5 Mb from the breakpoint; absent in GM12878 and GM24385. · Long-read multi-ome exposes inappropriate XCI of…; Discussion; Quantification of X chromosome and chromosome 13…; Figure legends

Methods, data and reproducibility

Permeabilized cells are labeled with the non-specific m6A methyltransferase Hia5 (Fiber-seq), and matched polyA cDNA is concatenated into ∼18 kb MAS-Seq arrays; libraries pooled 15:1 run on one PacBio Revio SMRT Cell and are then split by adapter for phased variant calling, assembly, CpG methylation, regulatory-element accessibility, and isoform analysis.

  • GIAB reference cell line GM12878 (HG001) from Coriell; pooled Fiber-seq gDNA and bulk MAS-Seq cDNA sequenced on one Revio SMRT Cell · human (GIAB HG001) · n = over 6 million reads per sample
  • GIAB reference cell line GM24385 (HG002) from Coriell; same pooled multi-ome design, used for variant, methylation, accessibility, and isoform benchmarks · human (GIAB HG002) · n = over 6 million reads per sample
  • Skin-biopsy fibroblasts from UDN participant UDN318336, a 9-month-old female with a de novo balanced X;13 translocation · human · n = single proband; 28x, 2,196,793 full-length transcripts
  • Retinal organoids differentiated from proband-derived iPSCs (Sendai reprogramming), harvested at day 145 for Fiber-seq on a PacBio Sequel II · human
  • Control fibroblasts (UND374570, UND359892) and AICS-0088 cl.83 iPSC-derived control retinal organoids used for Western blot comparison · human
  • GIAB v4.2.1 GRCh38 small-variant benchmarks (HG001, HG002) and NIST_SV_v0.6 HG002 SV benchmark lifted to GRCh38 · human (GIAB reference samples)
  • Orthogonal CpG methylation references for GIAB cells: EpiQC whole-genome bisulfite data and Epi2me Oxford Nanopore 60x data · human (GIAB reference samples)
  • ENCODE H3K27ac ChIP-seq and retinal DNase-seq tracks used to annotate the MAB21L1 locus · ENCFF365WHC, ENCFF522ANT, ENCFF962HWM, ENCFF064DIM
Methods & tools
Fiber-seq (Hia5 non-specific m6A-MTase labeling), MAS-Seq bulk cDNA concatenation (8-mer array) with Iso-Seq workflow in SMRT Link v12.0, DeepVariant 1.5.0, pbsv, HiPhase and meryl trio k-mer phasing, hifiasm and yak, jasmine and pb-CpG-tools, methylKit, fibertools-rs, FIRE (Fiber-seq Inferred Regulatory Element) pipeline, pbfusion, Western blot with Fiji quantification
Data & accessions
s3://stergachis-manuscript-data/2023/Vollger_et_al_long-read_multi-ome/GM12878_WGS.haplotagged.bams3://stergachis-manuscript-data/2023/Vollger_et_al_long-read_multi-ome/HG002_WGS.haplotagged.bams3://stergachis-manuscript-data/2023/Vollger_et_al_long-read_multi-ome/isoseq/analysis-NA12878_bulkMAS-Seq-204367-mapped.bams3://stergachis-manuscript-data/2023/Vollger_et_al_long-read_multi-ome/isoseq/analysis-HG002_bulkMAS-Seq-204368-mapped.bamhttps://s3-us-west-1.amazonaws.com/stergachis-manuscript-data/index.html?prefix=2023/Vollger_et_al_long-read_multi-ome/5mC/https://s3-us-west-2.amazonaws.com/stergachis-manuscript-data/2023/Vollger_et_al_long-read_multi-ome/UDN318336/trackHub/hub.txt
Software & environment
DeepVariant 1.5.0; SMRT Link v12.0; GIAB v4.2.1 GRCh38 and NIST_SV_v0.6 benchmarks; bedtools v2.30.0; Python v3.9.12; FIRE pipeline commit d095e67; SMRTbell prep kit 3.0; GRCh38 reference.
Assay conditions & compute
Each pooled library was sequenced on a single PacBio Revio SMRT Cell; retinal organoid Fiber-seq used a Sequel II system. No computational hardware, runtime, or memory figures are reported.
Reproducibility gaps
Some data restricted for participant confidentiality, with access terms from the corresponding author on request; Hia5 enzyme and plasmid only on request; FIRE pipeline cited as in preparation; versions missing for most tools.
Limitations (5)
  • Part of the study data is restricted to protect participant confidentiality, with access conditions available only on request from the corresponding author; the Hia5 enzyme and plasmid are likewise provided only on request. · stated by the authors
  • Heterozygous MAB21L1 loss of function is not known to cause disease, so the authors judge the MAB21L1 silencing unlikely to explain the proband's phenotype, which bears little resemblance to the biallelic MAB21L1 syndrome. · stated by the authors
  • Breakpoints for the earlier retinoblastoma case with an X;13 translocation, used as the comparator supporting the RB1 mechanism, were only estimated from its karyotype. · stated by the authors
  • Mechanisms rest on a single proband; elevated PDK3 protein comes from Western blots against one unrelated control organoid line, and the RB1 first-hit model is inferred from a ∼3% fibroblast accessibility imbalance rather than tested in retina or tumor tissue. · noted in this summary
  • No matched unpooled HiFi run is shown to support the claim that pooling leaves variant accuracy intact, F1 values are given only for GM24385, and the translocation bands differ between sections (p22.1;q14.1 vs p22.11;q13.3). · noted in this summary

Read this if you work on unsolved rare-disease cases, balanced translocations, or long-read multi-omics and want one PacBio run to link phased variants to methylation, chromatin, and transcript effects.

Citations
48
FWCI
27.43
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSanchis‐Jua '18Gorzynski '24Jammihal '26Jang '25Jensen '25Mostovoy '24
56% semantic matchGenomics
Alba Sanchis‐Juan, Jonathan C. Stephens, Courtney E. French+19 more·Genome Medicine·Dec 1, 2018· 174 citations