First chromosome-scale genome assemblies and comprehensive structural characterization of Tunisian durum wheat (Triticum turgidum subsp. durum) landraces Chili and Mahmoudi

Mapping the Genetic Blueprint of Tunisian Wheat

Researchers report a 125-fold improvement in genomic contiguity compared to previous durum wheat references. This breakthrough provides the first high-quality, chromosome-scale genetic maps for two vital Tunisian wheat landraces, Chili and Mahmoudi. These maps allow scientists to finally see the full structure of entire chromosomes. This visibility is essential for understanding how these plants survive extreme heat and drought.

Durum wheat is a cornerstone of global food security. It serves as the primary ingredient for pasta and couscous. Within the Mediterranean basin, Tunisia maintains a vital reservoir of "landraces"—locally adapted, traditional varieties. These varieties hold the genetic keys to climate resilience. Two such varieties, Chili and Mahmoudi, are prized for high protein content and arid-climate adaptation. However, no high-quality reference genome assemblies existed for these specific varieties until now.

Without a complete genomic blueprint, breeders are essentially flying blind. They cannot easily pinpoint the specific genes responsible for drought tolerance or grain quality. This makes it difficult to accelerate the development of crops that can withstand a warming planet.

The challenge of massive, repetitive genomes

Assembling a genome is like reconstructing a massive encyclopedia after it has been shredded. For simple organisms, this is straightforward. For durum wheat, it is a nightmare of complexity.

Wheat is a polyploid (an organism with multiple sets of chromosomes). Specifically, it has an AABB genome structure. This creates massive redundancy. Many genes appear in multiple, nearly identical copies across different sub-genomes. These extra copies are called homoeologs (related genes from different sub-genomes).

Wheat genomes are also enormous. They often exceed 10 gigabases (Gbp). They are packed with repetitive sequences. These repeats act like noise in a signal. They make it incredibly difficult for standard tools to determine exactly where a piece of DNA belongs.

Existing references, such as the Svevo v1 assembly, suffered from low contiguity (the length of continuous DNA stretches). Low contiguity means the genome is broken into millions of tiny, disconnected fragments. This makes it impossible to see the big picture of how genes are organized. As illustrates, previous iterations lacked the scale needed for a chromosome-level view.

A two-layer assembly and sketching strategy

The authors employed a specialized two-layer assembly pipeline to bridge the gap between small fragments and full chromosomes.

1. Long-read assembly: The team used PacBio HiFi long reads with the hifiasm assembler. Long reads are essential. They are long enough to span many repetitive regions that trip up shorter technologies.
2. Proximity scaffolding: To organize these reads, they used Illumina Hi-C data. Hi-C detects "proximity ligation" (physical closeness of DNA segments). This allows researchers to build "scaffolds" (long DNA chains) by seeing which parts touch one another.
3. K-mer sketching for assignment: A major hurdle is the 32-bit integer limit in common alignment tools. This limit prevents them from processing genomes larger than ~2.1 billion base pairs. To bypass this, the authors used sourmash. This uses "k-mer MinHash sketching" (creating a digital fingerprint of DNA sequences). This allows for rapid similarity estimation without massive computational power.
4. Alignment-guided resolution: Automated scaffolding often makes mistakes in polyploid wheat. It can accidentally fuse multiple chromosomes into one giant, incorrect scaffold. The authors used wfmash to align these "inflated" scaffolds against the Svevo v2 reference. They then split them at precise intervals to recover the 14 correct chromosomes.

This process is visualized in .

It shows how a massive, incorrectly joined scaffold was successfully decomposed into its rightful chromosomal components.

Achieving chromosome-scale precision

The results represent a significant jump in assembly quality. The authors report the Chili assembly spans 10.84 Gbp with a scaffold N50 of 756.2 Mbp. The Mahmoudi assembly spans 10.70 Gbp with a scaffold N50 of 756.8 Mbp.

In this context, "N50" is a metric of contiguity. It means that half of the total assembly is contained in scaffolds that are at least ~756 million base pairs long. Since a typical wheat chromosome is roughly 700–800 Mbp, these assemblies are effectively "chromosome-scale." This is a massive improvement over the Svevo v1 reference. That reference had a scaffold N50 of only 5.97 Mbp. That is a more than 125-fold increase in continuity.

The authors also measured biological completeness using BUSCO (a tool checking for essential, conserved genes). They found completeness scores of 99.4% for Chili and 99.3% for Mahmoudi. Additionally, using a reference-free tool called Merqury, the authors reported high base-level accuracy (QV scores of ~68). This ensures the individual letters of the genetic code are highly reliable.

Limits of the current blueprint

While these assemblies are a major milestone, the authors define several limitations.

First, the study focused on the physical structure of the DNA. It did not perform "annotation" (labeling functional genes and regulatory elements). The map is complete, but the instructions have not been translated yet.

Second, chromosome assignment relied on k-mer sketching rather than full whole-genome alignment. This was due to the software limits mentioned earlier. While they used an alignment-guided step to fix errors, they admit some uncertainty remains. Fine-scale ordering of repetitive regions might require further validation through genetic mapping. Finally, the study does not address all potential structural rearrangements in these landraces.

The verdict: A toolkit for democratization

If you are looking for a production-ready reference for North African durum wheat, the answer is yes. The authors have provided high-fidelity, chromosome-scale resources. These are ready for immediate use in genome-wide association studies (GWAS) and marker-assisted breeding.

Most importantly, the authors demonstrate that complexity does not require a massive supercomputer. They executed the entire workflow on the public Galaxy Europe platform. They proved that reference-quality polyploid assembly is achievable on consumer-grade hardware. It required as little as 15 GB of RAM. This effectively democratizes high-end genomics for researchers in resource-limited settings.

The assemblies and pseudomolecules are openly available via Zenodo (10.5281/zenodo.20366290).

Figures from the paper