Population-Scale Pangenome Research Solutions Using Long-Read Sequencing

De novo assembly, graph-based pangenome construction, structural variant discovery, and trait association analysis for population-scale genomics.

CD Genomics provides dedicated pangenome research solutions that combine long-read sequencing (PacBio HiFi and Oxford Nanopore) with comprehensive bioinformatics for population-scale genomic analysis. Unlike single-reference assembly projects, our pangenome service is designed for multi-sample cohort studies — from de novo genome assembly and graph-based pangenome construction through structural variant (SV), presence-absence variation (PAV), and copy-number variant (CNV) discovery, gene family classification, and trait association analysis. We support pangenome projects across plant, animal, microbial, and human species, with flexible project packages sized from pilot studies (5–10 samples) to large breeding and mechanism cohorts (200+ samples).

What Makes This Service Different

• Population-scale pangenome, not single-genome assembly. Our service is built for multi-sample cohort analysis — constructing graph-based pangenomes that capture the full spectrum of genetic diversity across a population, including core, dispensable, and private genomic content that single-reference approaches miss.
• SV/PAV/CNV-centric discovery framework. Long-read sequencing enables direct detection of structural variants, presence-absence variation, and copy-number variants at nucleotide resolution. We prioritise these variant classes because they are the largest contributors to functional genetic diversity and are systematically missed by short-read pangenome methods.
• Designed for trait association and breeding applications. Our pipeline connects pangenome variation to functional outcomes — through GWAS, candidate gene identification, selection sweep analysis, and functional enrichment — delivering biological insight rather than raw assemblies. This makes the service directly applicable to breeding programmes, evolutionary genomics, and disease-variant discovery.

For Research Use Only. Not for use in diagnostic procedures, clinical decision-making, personal health assessment, or therapeutic decision-making.

Why Long-Read Sequencing for Pangenome Analysis?

A single reference genome cannot represent the full genetic diversity of a species. Pangenome analysis addresses this limitation by incorporating genomic information from multiple individuals, constructing a graph or matrix that captures both shared (core) and variable (dispensable/private) genomic content. However, the resolution and biological value of any pangenome depend entirely on the sequencing technology used to build it.

Short-read sequencing (<150–300 bp) has been the predominant approach for pangenome studies, but it has fundamental blind spots. Structural variants larger than the read length, tandem repeat expansions, segmental duplications, and insertion sequences (including transposable elements and endogenous viral elements) are poorly resolved or completely invisible to short-read methods. Presence-absence variation in repetitive regions — a major source of functional diversity in plants and animals — is systematically underestimated. Studies comparing short-read and long-read pangenomes consistently report that short-read approaches miss 50–70% of structural variants and substantially underestimate the dispensable genome fraction.

Long-read sequencing overcomes these limitations. PacBio HiFi reads (>99.9% accuracy, 15–25 kb average length) provide single-nucleotide resolution across the full length of SVs and repeats, enabling direct genotyping of insertions, deletions, inversions, and translocations. Oxford Nanopore ultralong reads (100+ kb) span the largest repetitive elements and complex structural rearrangements, providing contiguity for de novo assembly and gap closure. When combined, these platforms deliver pangenome assemblies and variant call sets that capture the true extent of genetic diversity within a population — including the large-effect variants most likely to underlie phenotypic variation, disease susceptibility, and adaptive traits.

What We Deliver — From Assemblies to Biological Insight

Every pangenome project is built on a foundation of high-quality individual genome assemblies, from which we construct graph-based pangenomes, call structural variants, classify gene content, and perform trait association analyses. Our deliverable structure is designed to provide both raw data (for downstream custom analysis) and interpreted results (publication-ready figures and reports).

Recommended Project Packages

Project scoping depends on research objectives, genome complexity, and desired analytical depth. We recommend the following package configurations as starting points for discussion. Specific sample numbers, coverage levels, and analytical scope can be adjusted to match study requirements.

Reference benchmark. The above coverage recommendations and project designs are informed by published long-read pangenome studies including the 32-ecotype Arabidopsis thaliana pangenome built from PacBio HiFi data (Kang et al. 2023), the 53-individual Arab human pangenome using HiFi + ONT ultralong sequencing (Nassir et al. 2025), and the HPRC human pangenome draft (Liao et al. 2023). These studies demonstrate that long-read pangenome analysis at 10–60× HiFi coverage produces assemblies and variant call sets that substantially exceed the resolution achievable with short-read approaches.

Sample Requirements

Case Study: A Graph-Based Pangenome of 32 Arabidopsis thaliana Ecotypes Using PacBio HiFi Sequencing

To demonstrate the power of long-read pangenome analysis, we highlight the work of Kang et al. (2023), who constructed a graph-based pangenome of Arabidopsis thaliana from 32 ecotypes sequenced with PacBio HiFi long reads. This study, published in Nature Communications (CC BY 4.0), illustrates how long-read sequencing at population scale captures structural variation that is invisible to short-read methods and connects it to local adaptation.

1. Background

Arabidopsis thaliana is the premier plant model system, but existing pangenome studies relied primarily on short-read data that could not resolve large structural variants or repeat-rich regions. The ecotypes selected for this study span the species' global range, including relict populations from the Qinghai-Tibet Plateau (Tibet-0), Italy, and Morocco — representing ancestral lineages — and postglacial expansion ecotypes from across Eurasia.

2. Methods

The authors generated de novo assemblies for all 32 ecotypes using PacBio HiFi long-read sequencing (mean coverage depth sufficient for high-quality assembly). A graph-based pangenome was constructed using the Minigraph-Cactus pipeline, producing a variation graph with 468,168 nodes and 649,692 edges spanning 243.27 Mb of genomic sequence. Structural variants were called and genotyped across all ecotypes, with transposable element associations analysed for each SV class.

3. Results

Representative pangenome analysis results from Kang et al. (2023). Panels adapted from the original publication under CC BY 4.0. Left: Pangenome graph visualisation showing structural variant density across the A. thaliana genome. Right: PAV distribution across 32 ecotypes and identification of an HPCA1 promoter SV associated with alpine adaptation in the Tibet-0 ecotype.

Key Findings

• 61,332 structural variants (≥50 bp) identified across 32 ecotypes, overlapping 18,883 genes — demonstrating that SV burden in the A. thaliana genome is substantially higher than short-read pangenome estimates.
• Multiallelic SVs (29.65 Mb) were the largest category, highlighting the complexity of structural variation that graph-based pangenomes can capture but linear references cannot.
• ~22.69% of SVs correlated with transposable element insertions, with large SVs (>500 bp) predominantly TE-associated, implicating TE activity as a major driver of genome structural diversity.
• A 332 bp promoter insertion in the HPCA1 gene was identified in the Tibet-0 ecotype, enhancing gene expression and contributing to adaptation to high-altitude alpine environments — a demonstration of how pangenome-resolved SVs can be directly linked to local adaptation.
• Relict ecotypes harboured more unique SVs than postglacial expansion ecotypes, suggesting ancestral genetic diversity that has been lost during range expansion.

4. Conclusions

This study demonstrates that long-read-based pangenome analysis uncovers structural variation at a resolution and scale that short-read methods cannot achieve. The identification of an adaptive SV directly linked to environmental adaptation illustrates the biological and practical value of population-scale long-read pangenome projects for understanding the genetic basis of phenotypic variation.

Demo & Inquiries

Below are representative output formats from a long-read pangenome project. These AI-generated illustrations demonstrate the types of results delivered with each project, including pangenome graph visualisation, SV landscape summary, and association results.

1. Pangenome graph visualisation showing structural variant distribution across assembled genomes and the core/dispensable/private gene classification summary.

2. SV/PAV/CNV landscape summary including variant type distribution, size spectrum, and population frequency spectrum across the study cohort.

3. Trait association results (Advanced module) — Manhattan plot from SV-based GWAS and selection sweep landscape across the genome.

Representative pangenome analysis deliverables: pangenome graph visualisation, SV/PAV distribution across samples, and GWAS/selection sweep results. AI-generated representative data — not from a specific study.

To discuss your pangenome project requirements, sample numbers, species-specific considerations, or to request a detailed quotation, please contact our scientific team.