CD Genomics

06/09/2026

New in Nature Computational Science (2026): MetaSTAARlite — a scalable pipeline for rare-variant meta-analysis at biobank scale, without sharing individual-level data.

The problem it solves: UK Biobank, All of Us, and hospital biobanks each hold large WGS datasets. But data-sharing restrictions make direct pooling legally and logistically difficult. MetaSTAARlite works on summary statistics only — raw genotypes never leave each institution.

Key validation numbers: ✔ 190,110 UK Biobank WGS individuals across three cohorts ✔ Cross-validated with UK Biobank WES (446,933) + All of Us WGS (94,532) ✔ Results highly concordant with individual-level pooled analysis ✔ Compute, memory, and storage scale linearly — biobank-ready

Covers coding + noncoding regions with functional annotation integration.

Read the paper:

MetaSTAARlite provides a scalable and resource-efficient summary statistics-based pipeline for powerful, functionally informed rare-variant meta-analysis of biobank-scale sequencing data.

06/09/2026

New from Nature Reviews Methods Primers (2026): a comprehensive Primer on enzyme-tethering methods for in situ epigenomics — authored by the Henikoff lab and collaborators.

The key distinction from ChIP-seq: enzyme-tethering methods (CUT&RUN, CUT&Tag, DamID, RT&Tag) keep cells and nuclei intact during DNA modification. ChIP-seq dissolves chromatin first.

Why it matters: ✔ Lower background, higher resolution ✔ Single-cell and spatial applications (Spatial-CUT&Tag, scDam&T-seq) ✔ Works with low-input / rare samples ✔ FFPE archival tissue compatible (FFPE-CUTAC) ✔ Multi-factor + multi-omics in one cell

The Primer also covers time-resolved epigenome mapping (CUT&TIME, SLAM-RT&Tag) for tracking chromatin dynamics.

CD Genomics offers CUT&Tag and CUT&RUN as part of our epigenomics service portfolio — optimized for low-input and single-cell workflows.

Read the Primer:

Enzyme-tethering epigenomic methods modify DNA while cells and nuclei remain intact, making them particularly suitable for single-cell and spatial applications. In this Primer, Kami Ahmad et al. discuss the various approaches for enzyme-tethering analysis and major applications in biological researc...

06/09/2026

Standard amplicon-based mtDNA sequencing has two problems most researchers discover too late: PCR bias and loss of epigenetic marks.

CD Genomics' mtDNA Sequencing service offers three routes — matched to your research question:

🔬 Three sequencing options: ✔ Illumina (MiSeq/HiSeq PE150) — 100% genome coverage, no blind zones. SNV/InDel, heteroplasmy, circos output. Input: ≥0.4 µg gDNA. ✔ Mitochondrial capture sequencing — probe enrichment from blood or total DNA, no mitochondrial isolation needed. ✔ PacBio long-read, amplification-free — full-length linearized mtDNA, preserves base modifications. For epigenetics, phasing, and structural variants.

📊 Published: Current Biology (IF 9.5) n=10,442 patients. GWAS found TFAM + CDK6 regulate mtDNA levels. Major depressive disorder patients showed elevated heteroplasmy. Chronic stress induces heteroplasmy — mtDNA as a biological stress readout.

From mitochondrial disease to population genetics to toxicology — method, platform, and sample type all matched.

🔗 https://www.cd-genomics.com/mitochondrial-dna-mtdna-sequencing.html

06/09/2026

The exome is 1–2% of the genome. It contains >85% of known disease-causing variants.

That's why WES outperforms WGS on cost-per-finding for most germline and somatic research questions. But the method details still determine what you can actually detect.

CD Genomics Human/Mouse Whole Exome Sequencing:

🔬 Three human panels: ✔ Core Panel (~34.4 Mb CDS) — germline, population, translational @ 100–150× ✔ Inherited Disease Panel — 641 genes, OMIM/ClinVar/pathogenicity scoring @ ≥11 Gb ✔ Tumor Panel — TMB, MSI, HLA, fusion genes, FFPE-optimized @ ≥200×, 20 Gb

🐭 Mouse Standard Panel — ~38 Mb CDS, ≥100× at just 8 Gb. Built for transgenic/KO model validation.

⚡ Long-read exome available: Nanopore + PacBio HiFi panels for structural variants, phasing, and complex FFPE samples — not standard at most CROs.

🧬 Published: NAR Cancer 2025 — Exome + optical genome mapping in ccRCC. VHL/PBRM1/SETD2 mutations, 3p deletion, and global 5hmC loss identified in one integrated workflow.

Low-input protocol handles ≥50 ng starting DNA. Delivers annotated VCF to Excel report.

🔗 https://www.cd-genomics.com/human-mouse-exome-sequencing.html

06/09/2026

TCR-seq result quality is determined at library prep — not sequencing depth.

Most multiplex PCR (mPCR) protocols amplify abundant clones preferentially. Rare clones — the ones that often matter most in tumor immunity and minimal residual disease — get systematically underrepresented.

CD Genomics uses 5' RACE for TCR library construction: one primer pair per cycle, full-length TCR capture (CDR1 + CDR2 + CDR3), lower bias, more accurate diversity readout.

🔬 What 5' RACE gives you: ✔ Full-length variable region — CDR1/2 for MHC affinity, CDR3 for antigen specificity ✔ Lower PCR bias vs. multiplex PCR ✔ Sequencing depth matched to question: 6G (structural) → 10-20G (rare clone tracking) ✔ Compatible with RNA, DNA, PBMC, blood, and tissue

📊 Published: Cancer Science 2024 (IF 5.7) HCC patients across BCLC stages — TCR profiling in tumor tissue, peritumoral tissue, and PBMC. Key finding: BCLC-C stage patients showed increased TCR clonality and decreased diversity. Higher clonality correlated with lower recurrence risk and better prognosis.

TCR repertoire as a prognostic tool — with the data to back it up.

🔗 https://www.cd-genomics.com/tcr-seq.html

06/04/2026

The most common Hi-C budget mistake: optimizing for resolution, not for your question.

When researchers ask "how much Hi-C depth do I need?", the instinct is to anchor on the highest resolution tier — 5 kb, 1 kb, whatever the benchmark paper used. That instinct is often wrong.

Resolution is an output. Your biological question is the input. Planning depth in the wrong order wastes budget and, worse, produces a dataset that can't actually support the conclusion you need.

Dr. Yang H., Senior Scientist at CD Genomics, breaks down the reasoning in our latest resource article. Three principles stood out:

1. There is no universal correct depth. The same read count that comfortably calls A/B compartments will severely underpow­er loop detection. "High resolution Hi-C" is not one experiment — it's a family of experiments with very different cost structures.

2. "Maximum resolution" is often the wrong budget target. If your question is genome-wide structural comparison, TAD-level context mapping, or a scaffold for targeted follow-up, a moderate depth is entirely defensible — and often better than overcounting a small region.

3. Phased planning reduces financial risk. Phase 1: Generate enough data for QC, complexity assessment, and signal review at your target scale. Decision gate: Will deeper sequencing materially change the conclusion? Phase 2: Only proceed if the evidence supports it.

The piece also covers why library complexity is the real ceiling on depth (additional sequencing past saturation only adds duplicates, not resolution), and when switching to Capture Hi-C or Micro-C is more cost-efficient than simply sequencing deeper.

Worth bookmarking if you're designing a 3D genomics project or advising collaborators on scope.

👉 Full article by Dr. Yang H.: https://www.cd-genomics.com/3d-genomics/resource/hi-c-budget-planning-sequencing-depth.html

06/04/2026

A common mistake in HiChIP study design: choosing an antibody target because it's familiar, not because it matches the regulatory question.

The result is technically clean data that can't answer the hypothesis. That's not a sequencing failure — it's an anchor selection failure.

The anchor is the lens. If the lens doesn't match the mechanism you're testing, HiChIP can return a large, well-QC'd dataset with almost no decision value.

Our senior scientist Dr. Yang H. breaks down why this happens and how to avoid it in a new resource article. The short version:

Three things a biologically strong HiChIP anchor needs to satisfy:

Mechanism match — Are you testing enhancer-driven activation or promoter-centric connectivity? Those require different anchors. Using a broadly enriched mark when your hypothesis is enhancer-specific gives you interpretable-looking output for the wrong layer.

Meaningful simplification — The right anchor should help you prioritize interactions that are plausible for your model, not merely abundant. A richer interaction map is not always a more informative one.

Defensible validation path — Can you sketch in advance what a "convincing" result would look like, and what orthogonal data would support it? If not, the anchor may not be specific enough for the question.

The five traps most HiChIP projects fall into:

Selecting the target before defining the regulatory claim
Assuming stronger enrichment = stronger evidence
Expecting a single anchor to answer every regulatory question
Under-planning how the output connects to downstream validation
Defaulting to the most-cited target without checking mechanism fit
When the anchor is wrong, the deliverable isn't just suboptimal — it's actively harder to interpret than a simpler experiment would have been.

Read the full article: https://www.cd-genomics.com/3d-genomics/resource/hichip-target-selection-wrong-anchor-weakens-study.html

06/04/2026

Hi-C can tell you that locus A contacts locus B, and that locus B contacts locus C.

What it cannot tell you is whether A, B, and C are simultaneously in the same physical complex — or simply sharing a neighborhood at different moments. That distinction is the difference between an enhancer hub and a coincidence. Standard paired-end sequencing is structurally incapable of answering it.

HiPore-C resolves this directly.

By sequencing intact concatemers on Oxford Nanopore PromethION, each read encodes a complete multi-way contact event — three or more restriction fragments from a single ligation molecule, captured simultaneously. There's no inference. There's no statistical reconstruction. The co-occupancy is in the read.

What HiPore-C delivers that paired Hi-C cannot:

→ Higher-order contacts (≥3 loci per read) — enhancer hubs, LCR multi-gene contacts, and multi-way regulatory complexes captured as single molecules → Complex SV resolution — N50 >15 kb reads cross repetitive regions, centromeres, and SV breakpoints that short reads cannot span → Allele-specific interactions — heterozygous SNPs across a single long molecule enable haplotype-resolved chromatin contacts without computational phasing overhead → Direct V2G evidence — physically linking a risk variant, its enhancer, and its target promoter in one molecule; not a statistical prediction

The benchmark: HiPore-C at the β-globin locus (Nat Commun. 2023) — the locus control region (LCR) was found to simultaneously contact ≥2 globin genes in a single molecule, confirming a true enhancer hub rather than sequential pairwise interactions. Multi-way contact frequency exceeded pairwise model predictions, demonstrating that Hi-C structurally underestimates hub complexity.

Recommended depth: 50–100 Gb per sample. Analysis via PPL-Toolbox (Pore-C Pipeline-Toolbox), delivering contact matrices compatible with standard Hi-C viewers plus dedicated higher-order interaction tables.

Best fit for: V2G prioritization in drug target discovery, cancer structural variant 3D impact mapping, imprinting and X-inactivation allele-resolved studies, complex genome scaffolding.

Service details: https://www.cd-genomics.com/3d-genomics/services/hipore-c.html

06/04/2026

The resolution floor of standard Hi-C is set by its restriction enzyme.

Cut with DpnII, and your smallest detectable fragment is ~4 kb. That means any chromatin loop shorter than that — the short-range enhancer-promoter contacts that drive transcription in the 1–20 kb window — sits below the detection threshold. You're looking at chromosomal architecture with a map that can't read street-level.

Micro-C replaces the restriction enzyme with MNase.

Micrococcal nuclease digests exposed linker DNA uniformly, generating fragments centered on single nucleosomes (~150 bp) and di-nucleosomes (~300 bp). There's no sequence motif bias, no AT-rich deserts left uncut, no genomic blind spots. Every ligation product captured reflects contacts between adjacent nucleosomes — not between restriction fragments 4 kb apart.

What this unlocks:

→ ~150 bp resolution — one order of magnitude finer than standard Hi-C → Short-range loops (

06/04/2026

Every Hi-C experiment gives you pairwise contacts. Two loci per read, every time.

That's been the fundamental limitation of standard chromatin conformation capture for 20 years: you can see whether locus A is near locus B, but you can't determine whether A, B, and C are simultaneously in the same physical complex — or just sharing a neighborhood at different times.

Pore-C changes the unit of measurement.

By sequencing intact concatemers on Oxford Nanopore — without fragmenting the ligation product — each read directly encodes multi-way contacts: three, four, sometimes more restriction fragments captured in a single molecule. The difference matters enormously when you're asking regulatory questions. Whether a risk variant, an enhancer, and a promoter form a co-occupying hub is not answerable from pairwise data alone.

What the platform delivers that standard Hi-C cannot:

→ Higher-order contacts — multi-way interaction tables alongside standard paired format (Juicebox / HiGlass compatible) → Native CpG methylation — base modification detected simultaneously from the same raw signal, no bisulfite treatment needed → Long-range scaffolding — reads averaging 15–30 kb (some exceeding 100 kb) bridge repetitive regions, centromeres, and telomeres that short reads cannot resolve → Chromosome-scale haplotype phasing — heterozygous SNPs + methylation imprints across a single long read enable allele-resolved assembly

A 2025 study using Pore-C with C-Phasing assembled an ultra-complex polyploid plant genome — previously intractable with PacBio contigs + short-read Hi-C — into chromosome-scale, haplotype-resolved scaffolds with >99% accuracy. Multi-way contacts bridged centromeric repeats that had left contigs isolated.

Standard deliverables: Raw .fastq with methylation tags, sorted .bam, .hic / .mcool contact matrices, multi-way .pairs tables, QC PDF with N50, concatemericity, cis/trans ratio, methylation detection rate.

If your project involves complex genome assembly, structural variant validation, enhancer-promoter hub analysis, or allele-resolved chromatin mapping — this is worth a closer look.

Service details: https://www.cd-genomics.com/3d-genomics/services/pore-c.html