Sequencing Revolution Races to Rescue the World’s Fruits and Vegetables

Few corners of agriculture are as biologically stubborn as the world’s horticultural crops. Fruits, vegetables and ornamentals feed humanity with the vitamins, minerals, fibre and phytochemicals that staple cereals simply cannot supply, yet they are notoriously hard to improve through conventional breeding. Perennial fruit and nut trees can take eight to ten years or more to complete a single breeding cycle because of long juvenile phases. Many species, including banana, potato, sweet potato and strawberry, are highly heterozygous, self-incompatible, vegetatively propagated or polyploid, which erodes the predictive power of pedigree-based selection and makes it difficult to fix desirable alleles without dragging along linked deleterious ones. A sweeping new review published in Discover Plants argues that next-generation sequencing has dismantled many of these barriers, but it also delivers a blunt warning: the real bottleneck is no longer generating sequence data, but converting it into deployable breeding decisions.

The review, led by Khwairakpam Rozerto of Nagaland University with colleagues in India, synthesises the full sequencing-to-breeding pipeline for horticultural species and positions itself deliberately against earlier surveys of plant genomics. Its central thesis is that the fastest-moving technologies, from telomere-to-telomere assemblies to artificial-intelligence-assisted genomic prediction, are unevenly represented in the horticultural literature, and that the practical constraint for most breeding programmes, particularly in the Global South, is downstream bioinformatics, data storage and functional-validation capacity rather than access to a sequencer. The authors searched PubMed, Web of Science, Scopus and Google Scholar with no lower date limit, prioritising records from January 2023 to June 2026 for claims about the current state of any technology, and they are explicit that this is a narrative expert synthesis rather than a formal systematic review.

At the heart of the analysis is a comparative evaluation of sequencing platforms. Illumina sequencing-by-synthesis remains the workhorse of horticultural genomics, delivering per-base accuracy above 99.9 percent at low cost, which makes it ideal for SNP genotyping, resequencing and RNA-seq at population scale. Its weakness is read length, typically 100 to 300 base pairs, which frustrates the resolution of repetitive regions, structural variants and haplotype phase, precisely the features that dominate the heterozygous, repeat-rich genomes of fruit trees. Long-read platforms have closed much of the gap: PacBio’s HiFi chemistry now yields 10 to 20 kilobase reads at greater than 99.9 percent accuracy, while Oxford Nanopore’s R10.4-class flow cells offer ultra-long reads exceeding 100 kilobases, real-time analysis, field portability and native detection of DNA base modifications without a separate bisulfite step. In practice, the review notes, most recent high-quality horticultural assemblies combine platforms, using long reads for contiguity, short reads for polishing and Hi-C chromatin-contact data for chromosome-scale scaffolding.

The payoff of this hybrid strategy is visible in the new generation of reference genomes. Telomere-to-telomere assemblies, gap-free sequences spanning every chromosome from centromere to telomere, have now been achieved for carrot, octoploid strawberry, apricot, apple, pear, melon, grapevine and mango, several with BUSCO completeness exceeding 98 percent. In mango, paired T2T assemblies of the cultivar Kensington Pride and its anthracnose-resistant wild relative Mangifera laurina revealed presence-absence variation in fruit-colour and aroma-biosynthesis genes, directly implicating candidate genes for disease resistance. Haplotype-resolved assembly, which reconstructs each parental haplotype separately rather than collapsing them into a single consensus, is proving transformative for clonally propagated crops such as grape, apple and kiwifruit; phased T2T genomes of apple have exposed substantial allelic divergence and expansion of resistance-gene families during domestication that a collapsed reference would mask entirely. For polyploid and highly heterozygous species, the authors argue, these approaches are not a technical refinement but a precondition for reliable genetic discovery.

Equally consequential is the shift from single reference genomes to pan-genomics. Any individual genome lacks genes and haplotypes present in other accessions, so a single reference, however complete, cannot represent a species’ full diversity. Pan-genomes partition this diversity into a core genome shared by nearly all accessions and a dispensable genome enriched for disease resistance, stress tolerance and other adaptive traits, characterised through presence-absence variation, structural variation and copy-number variation. Graph-based pan-genomes encode this variation as nodes and edges, with each path through the graph corresponding to a haplotype, and have been built for kiwifruit and Brassica species. A six-genome graph pan-genome of Brassica oleracea morphotypes, spanning cabbage, broccoli, cauliflower and kale, identified more than 200 structural-variant hotspots underlying leaf-versus-stem domestication and head formation. In tomato, pan-genome analysis of 725 accessions uncovered genes tied to domestication and fruit quality that the reference genome had missed, while a phased potato pan-genome has already been applied to haplotype-based design of hybrid potato, linking pan-genomic variation directly to a breeding strategy.

Transcriptomics is undergoing its own resolution revolution. Bulk RNA-seq reports average expression across all cells in a tissue, obscuring cell-type-specific and spatially restricted patterns that often carry the biology. Long-read transcriptomics with PacBio Iso-Seq and Nanopore direct RNA sequencing captures full-length transcripts, resolving alternative splicing and fusion transcripts without ambiguous short-read reconstruction. Single-cell and single-nucleus RNA sequencing, still concentrated in model species but expanding into strawberry and lychee, resolves expression at individual-cell resolution. Spatial transcriptomics retains positional context, and combined single-cell and spatial approaches have traced the spatiotemporal trajectory of fruit senescence from exocarp to mesocarp in pitaya. Perhaps most striking is the potato story: single-cell spatiotemporal profiling of leaves infected with the late-blight pathogen Phytophthora infestans distinguished directly targeted cells from their immediate neighbours and resolved distinct, coordinated defence programmes between the two, a distinction invisible to bulk sampling of infected tissue.

Epigenomics is emerging as a complementary lever. DNA methylation in CG, CHG and CHH contexts influences fruit ripening, stress tolerance, flowering time and disease resistance, and whole-genome bisulfite sequencing now makes genome-wide methylation profiling routine. Dysfunction of a DNA demethylase delays ripening in tomato, and methylation inhibitors alter ripening across multiple fruit species, suggesting epigenetic regulation of ripening is a shared feature of economically important fruits. Chromatin-accessibility profiling with ATAC-seq has identified stage-specific accessibility changes coinciding with sugar accumulation during apple fruit development and anthocyanin-related colour development in mulberry, linking specific transcription-factor-binding elements to the genes they regulate. CRISPR/dCas9-based epigenome editing allows targeted adjustment of methylation patterns without altering the genetic code, offering a potential non-transgenic route to modifying stress tolerance and quality traits, though the review cautions that whether stress-induced marks persist across generations of clonal propagation remains an open question.

On the genotyping front, the review dissects strategies often conflated in the literature. Marker-assisted selection tracks one or a few major-effect loci; genome-wide association studies are discovery-oriented; and genomic selection fits prediction models across the entire genome to generate genomic estimated breeding values, which is especially powerful for polygenic traits and perennial tree crops where seedling-stage selection can shave years off the breeding cycle. A low-coverage strategy called skim sequencing is highlighted as a pragmatic middle ground: in eggplant, where no SNP array existed, 3-fold coverage sequencing genotyped a 325-line MAGIC population at roughly 40-fold higher marker density than a prior targeted-capture approach, enabling GWAS mapping of root-architecture QTLs. Genome editing completes the pipeline, with base editing, prime editing and multiplexed CRISPR/Cas12 systems shortening the path from gene discovery to trait deployment; in citrus, ribonucleoprotein delivery of Cas12a produced ten transgene-free edited sweet orange lines, four carrying biallelic mutations at all three target sites of the canker-susceptibility gene CsLOB1.

Yet the review’s most sobering contribution is its accounting of what still blocks translation. Genome complexity, heterozygosity, polyploidy and repetitive DNA, complicates assembly and biases variant calling, while reduced-representation genotyping systematically under-calls heterozygotes through allelic dropout. Deep-learning variant callers such as DeepVariant, benchmarked on strawberry and tetraploid potato, missed roughly 60 percent of single-copy alternative-allele sites in potato, underscoring that tools trained on human and model genomes do not transfer cleanly to polyploid crops. Functional validation lags because horticultural species are often recalcitrant to transformation and perennials impose multi-year juvenile phases, so candidate genes accumulate faster than they can be confirmed. Genomic resources remain concentrated in tomato, apple, grapevine, citrus and potato, leaving orphan and underutilised crops, often stress-tolerant and nutritionally distinctive and grown in the regions least able to fund reference projects, excluded not from sequencing itself but from the analytical layer where sequence becomes breeding value. The authors’ prescription is a framework that matches each separable constraint, from assembly to validation to infrastructure, to the specific technology most likely to relieve it, and a clear priority: invest in bioinformatics capacity, phenotyping and validation throughput, because further reductions in sequencing cost will no longer, on their own, put better fruit and vegetables in the world’s markets.

Subject of Research: Applications of next-generation sequencing technologies in the genetic and genomic characterisation and breeding of horticultural crops

Article Title: Next-generation sequencing for genetic and genomic characterization of horticultural crops

Article References: Next-generation sequencing for genetic and genomic characterization of horticultural crops. (n.d.). https://doi.org/10.1007/s44372-026-00935-z

Image Credits: AI Generated

DOI: 10.1007/s44372-026-00935-z

Keywords: next-generation sequencing, horticultural crops, genome assembly, telomere-to-telomere, pan-genomics, genomic selection, CRISPR, epigenomics, transcriptomics, plant breeding, polyploidy, bioinformatics