Exploring Pan-Genomes of Crops: Genomic Resources and Tools for Unraveling Gene Evolution, Function, and Adaptation

1. Introduction

In recent years, advancements in affordable sequencing platforms and computational resources have helped to generate reference-quality genome assemblies for multiple crop accessions/varieties belonging to a single species. Thus, crop genomics has transitioned to the pan-genomic era. This shift has been made possible by advances in both low-cost sequencing (i.e., short-read data Illumina sequencing, long-read PacBio sequencing with low errors, Nanopore sequencing) and bioinformatics tools employed for genomic data processing and genome assemblies (for excellent reviews, see [1,2,3,4,5,6,7]. Whole-genome assembly-based pangenomes have been reported for a few plant species, including rice [8,9,10], barley [11], wheat [12], maize [13], soybean [14], a wild relative of soybean Glycine soja [15], and brassica [16]. In particular, PacBio HiFi reads proved very useful for high-quality polyploid genome assemblies for peanut [17], wheat [18], oilseed [16], strawberry [19] and potato [20].

The intraspecies genome comparisons of crops suggest extensive structural variation across diverse genotypes that affect both the genomic contents and plant function [14,21,22,23]. The structural variations (SVs) among genotypes of the same species include insertion, deletions (indels), and translocation of the small or large genomic regions that further cause presence-absence variation (PAV), copy number variation (CNV), chromosomal rearrangements, repetitive sequences (i.e., tandem gene duplications, repetitive sequences in non-coding regions of the genome, transposable elements, centromere repeats, etc.). The conserved genes present in all cultivars/genotypes/subspecies/strains within a species constitute the “core” genome, and the variable genes represent as the “dispensable” or “accessory” genome [2,24]. The accessory genome consists of the “shell genes” (found in most cultivars within a species) and the “cloud genes “ (present in only a small fraction of cultivars of a same species) as shown in Figure 1.

Thus, a single reference genome represents only a fraction of the species-wide genomic space, and a pan-genome represents species-wide genomic space [2,13,15,25,26]. Often, the pangenomes encompass genes found in crop wild relatives and ancestral species [27,28,29,30,31]. We like to note here that many useful genes which were lost during the crop domestication and extensive plant breeding [32,33,34] may be found in the “dispensable”/accessory genome of any crop species which are needed for adaptation to a range of biotic and abiotic stress conditions, and improving physiology, plant structure, yield and nutrient quality of crops [15,33,35,36,37,38,39]. Thus, the availability of plant pangenomes allows plant researchers and breeders to explore important candidate genes for improving yield and nutritional quality, and resilience of crops to a range of climatic conditions and pathogens. For instance, a few comparative genomic studies have revealed that gene amplification plays a vital role in disease resistance, abiotic stress tolerance, and other agronomic traits associated with plant development, architecture, yield, etc. [40,41,42,43,44,45,46,47,48,49]. The high-quality pangenomes also make it possible to study previously inaccessible regions of the eukaryotic genomes, including centromeres, long heterochromatic blocks, and rDNA regions, etc. that exhibit low recombination, and provide new insights into crop genome evolution [50].

Recently, many excellent reviews have been published on plant pan-genomes (see [1,2,51,52,53,54]. Here, we review the current strategies for constructing and visualizing crop pan-genome data, public genomic portals hosting pan-genes, pan-genome data, pan-genome browsers, and associated analytical tools. Furthermore, we highlight a few studies that have exploited a pan-genomic approach to discovering candidate genes associated with important agronomic traits. We also discuss the potential of pan-genome-driven translational research.

2. Pangenome Construction, Visualization, and Data Analysis Tools

The first step in setting up a pangenome infrastructure is the selection of a diverse set of representative genotypes for sequence assembly that capture as many genetic variants as possible with a limited panel of genotypes [14,21,55,56]. The second step is sequencing of individual genomes. The third step is assembly and construction of pan-genome. Several popular approaches are implemented for pan-genome assembly (for detailed reviews, see [1,57,58,59]. Here we briefly describe the basic tenets of three popular methods (see Figure 2).

The first approach uses a high-quality reference genome for mapping sequence reads generated from all other genotypes. Iterative refinement allows for a progressive improvement of the assembly with additional data. This strategy can minimize errors by exploiting the information from a high-quality reference genome and limiting the coordinate consolidation issue. However, this method requires the availability of a high-quality reference genome, which may not be available for all species or strains. Secondly, it is sensitive to misalignment errors or inaccuracies in the reference genome, which can potentially propagate errors throughout the assembly. Likewise, bias towards the reference genome may limit the detection of novel or divergent sequences.

In species where a reference is unavailable, de novo assembly of individual genotypes is generated, and then assembled contigs are mapped to each other [27]. The de-novo genome assemblies have become a method of choice due to the advances in long-read sequencing and the availability of fast algorithms for aligning long-reads to call structural variants [7].

Conceptually, de novo assembly of multiple high-quality reference genome sequences and their comparison by pair-wise sequence alignment is arguably the most powerful and accurate approach to detect all types of sequence variants at base-level resolution, including novel genomic elements and rearrangements. However, generating assemblies of polyploid plant genomes is still a challenge, and current methods are limited in detecting and phasing heterozygous structural variants, erroneously produce chimeric contigs joining different haplotypes or ignore alternative haplotypes [60,61]. This approach is time-intensive and requires significant computational and bioinformatic resources, especially for large genomes and complex variations. Repeat resolution can be challenging, leading to fragmented assemblies, as it relies heavily on sequencing depth to overcome repetitive regions and complex variations.

The third graph-based approach allows the addition of any variant to the reference as a node at the genomic location where it is discovered, and then haplotypes are associated with one of the reference genomes used to build the graph. Reads are then realigned to this genome leading to more accurate mapping. This method can accommodate new genomic data through iterative refinement, allowing for continuous improvement of the pangenome assembly. However, graph construction and traversal can be computationally intensive, especially for large and diverse pangenomes, and requires substantial computational resources. Typically, graph complexity increases with the addition of more genomes, potentially impacting scalability, and computational efficiency. Nonetheless, the graph-based pan-genomes can represent complex variations, including structural variants and large-scale rearrangements, facilitating identification of shared and unique genomic regions among individuals or strains, and aiding in excellent visualization of pan-genomes. A conceptual visualization of the pan-genome is shown in Figure 3.

Recently, Shang et al., 2022 [21] have constructed ‘Super Pan-genome of rice’ containing high-quality assemblies of 251 rice genomes including 202 accessions of domesticated Asian rice Oryza sativa, 28 accession of a Oryza rufipogon (the wild ancestor of O. sativa), 11 accessions of domesticated African rice Oryza glaberrima, and ten accessions of Oryza barthii (the wild ancestor of O. glaberrima) using de novo long-read assembly and a graph-based approach. A reference-free whole-genome multiple sequence alignment for the 251 rice accessions that can be visualized using any assembly as the reference with the Rice Super Pan-genome Information Resource Database (http://www.ricesuperpir.com). This fully annotated pan-genome graph, visualized using the JBrowse genome browser, also integrated the SVs, gene annotations, transposable elements annotations, pan-genome graph, and BLAST tools [21].

A few excellent reviews have previously been published on the development of computational tools for pan-genome visualizations [57,58,62,63]. We note here that genome sequencing technologies and assembly algorithms, as well as the pan-genome visualization tools, are rapidly evolving to achieve high accuracy and are being complemented by additional independent mapping approaches, such as optical maps and Hi-C to validate structural variant calls (i.e., inversions and translocations). It is important to acknowledge that the details and outcomes of each method may vary based on the specific pangenome assembly tools, parameters, and characteristics of the genomic data employed in the process [64,65]. Here we compiled a summary of the latest representative tools in Table 1. It is crucial to recognize that the development of pan-genome tools is an active field, and the list could not be exhaustive.

3. A Survey of Crop Pan-genomes and Genomic Resources

With technical advances and affordability of the sequencing and assembly of the genomes, we are experiencing a deluge of big data in biology. The plant research community now faces a bigger challenge of making the genomic data findable, accessible, interoperable, and reusable (FAIR) [111,112]. Public databases and genomic resources play a crucial role in making genomic data FAIR and provide tools for visualization of genomic, transcriptomic, proteomic and metabolomic data [113,114,115,116,117,118,119,120]. Furthermore, the secondary knowledgebases synthesize and curate knowledge graphs providing information for gene-gene interactions, metabolic networks, and pathways, and provide the tools for analyses of user’s data in the context of plant genome browser or pathways [90,97,114,115,119,121,122,123,124,125,126,127,128,129,130]. Currently, a substantial number of plant genome browsers, amounting to a few hundred, can be accessed through platforms like Plant Ensembl [120], Phytozome [131], and various clade-specific community databases (for a recent review see [132]). However, the availability of plant pangenome portals is limited and experiencing slow growth. To provide an overview of the current state of crop pan-genomic research, resources, and portals, we compiled Table 2.

It is clear from the Table 2 that at present crop pan-genome research is at its early stage. The pan-genome browsers are available for a few crops, and thus, most of the data is not supported for user-friendly query, visualization and supporting the analysis of the user’s data. However, the very few platforms and genomic databases that support pan-gene analysis in phylogenomic context and support the user’s query show the potential of pan-genome data for answering both basic research and translational applications of this knowledge for crop improvement. Here, we highlight an example of visualizing the pan-gene data for TILLER ANGLE CONTROL 1 (TAC1) transcription factor from various accessions of cultivated rice O. sativa, other members of Oryza genus at Gramene oryza pansite (https://oryza.gramene.org). OsTAC1 is induced by gravistimulation and promotes horizontal shoot growth by negatively regulating shoot gravitropism [160]. Thus, it is involved in regulation of tiller angle and modulating plant architectural traits of agronomic tremendous importance. A comparison of TAC1 protein sequences shows a significant variability at the carboxy terminal between the domesticated rice cultivars of japonica and indica accessions (Figure 4A). Indeed, a previously published study has shown that a point mutation in OsTAC1 gene at the 3’-splicing site of the 1.5-kb intron (‘GGGA’) in japonica rice accessions caused reduction in the expression of this gene leading to a smaller tiller angle. This trait was selected in the japonica rice accessions. In contrast, wild rice accessions and indica rice accessions, which have large tiller angle, contain ‘AGGA’ sequence at 3’-splicing site of the 1.5-kb intron [161]. OsTAC 1 gene is on chromosome 9 in rice genome and shows high conservation across all rice accession (Figure 4B). However, we also see that genomic neighborhood of the TAC1 on left-hand side is not much conserved and it remains to be explored if any of these genes are involved in functional adaptation of rice accessions.

Beside Gramene database, a few more pan-genome portals provide a similar view of pan-genes visualization. The other portal host different views and analysis tools. For example, the Banana Genome Hub uses Panache platform for visualization of pan-genomic data on Musaceae. However, in general the well-established crop genome portals supports users in exploring genes and gene families, chromosome structures, synteny, SVs, gene expression patterns, SNP markers, etc.

4. Plant pan-genomics-driven insights for understanding the basis of agronomic traits

Cereal crops have been the prime subject of agriculture research. Thus, we have matured genomic resources, enriched genome annotations, and genotype and sequence data facilitating comparative genomics and pan-genomics studies (See Table 2). The focus of all categories of genomic research on cereal crops aims to increase the grain yield or plant developmental, physiological, and architectural traits that can support the higher yield. For example, Wang et al. (2022) [109] used rice pan-genome to identify GW5 genes associated with the trait ‘thousand-grain weight’ (TGW), and a novel locus qPH8-1 involved in the regulation of plant height. In another study by Shang et al., 2022 [21] a super pan-genome of rice was constructed helped to identify genetic variants associated with submergence tolerance, seed shattering, and plant architecture [21]. Many important studies have been published in maize, rice, wheat and are being implemented for their improvement.

Notably, more investments are being made in the genomic and pan-genomic research of minor cereal crops, including sorghum and millets suitable for growing in diverse and marginal lands (see Table 2). These crops have a high degree of in-built tolerance for mitigating the impact of harsh environments and resistance against many pests and pathogens. For example, foxtail millet (Setaria italica), is a model plant for studying C4 photosynthesis and developing climate-resilient crops. A pan-genomic study of foxtail millet identified an important genetic variation in the promoter region of SiGW3 that is associated with yield improvements [22]. Another study identified 13 marker-trait associations using proso millet (Panicum miliaceum L.) pan-genome [162]. Similarly, Yan et al. (2023) recently constructed a pearl millet pan-genome that identified over 400,000 genomic structural variants (SVs) and provided insights into heat tolerance. This study identified the RWP-RK gene conferring enhanced heat tolerance [55]. Another group of previously understudied crops is legumes that has gained from genomic and pan-genomic research and breeding efforts (see Table 2). For example, pigeon pea is an important orphan crop mainly grown by smallholder farmers in the tropics and subtropical regions of the world. It has an in-built tolerance for drought stress and is very productive in marginal land with small inputs. A pan-genome study of pigeon pea identified 225 SNPs associated with nine agronomically important traits. These associations will aid pigeon pea germplasm improvement for breeding legumes [154]. Mung bean is an excellent plant-based source of protein and is grown in temperate, subtropical, and tropical regions. In a study, Liu et al. (2022) analyzed 217 mung bean accessions and discovered many novel genes associated with agronomic traits, for example, an SNP in the candidate genes SWEET10 homolog (jg24043) associated with crude starch content; NRT1/PTR FAMILY 2.13 gene for pod length; a homolog of WUSCHEL-family homeobox gene associated with yield; and a PAV event in a multi-gene locus associated with color-related traits [163].

Pan-genome studies have become increasingly important for understanding the genetic diversity of major crops from tropical regions, including cassava (Manihot esculenta) and banana (Musa spp.). These crops are a vital component of diverse ecosystems and play essential roles in the livelihoods of local communities. Insights gained from genomic and pan-genomic research are being deemed vital for improving banana for resistance against Fusarium wilt to safeguard the global banana industry [99,164]. Similarly, the pan-genome of cassava (Manihot esculenta), a staple crop for millions of people in Africa, is helping to score genomic variations, particularly in genes associated with disease resistance and starch biosynthesis [151]. More studies are being performed on important fruit and vegetables (i.e., banana, apple, tomato, melon, citrus, and grape) and oilseed crops (i.e., Brassica, soybean, sunflower, etc.) listed in Table 2. A few studies have uncovered new genes and rare alleles that regulate secondary metabolites associated with color and flavor [46,49,165]; pathogen resistance [43,47,56,79,145,165,166,167]; and abiotic stress tolerance [41,42,55]. These findings foster understanding of the genetic basis of diverse traits in the domesticated crops and offer promising prospects for the introduction of candidate genes (for disease resistance breeding and quality traits) through precise genome editing into elite cultivars.

In conclusion, the acquisition of pangenome holds immense potential for pursuing fundamental questions, especially for climate-change resilient breeding and overall crop improvement. Utilizing pangenomes enables researchers to (1) Identify gains and losses of genetic regions and structural variations strongly associated with desirable fitness phenotypes such as abiotic stress and disease tolerance, growth and development, yield, biomass, and performance. (2) Identify gains and losses in protein-coding regions and/or epigenetic features between crops and closely related species. (3) Identify pathways, gene networks, expression profiles, and transcript isoforms that are enriched for given major and minor quantitative trait loci (QTLs) with desired phenotypes and adaptation traits. (4) Project functional and phenotype homologs from a well-studied species onto a new/less-studied species through whole genome comparisons and synteny. (5) Advance plant breeding efforts by mapping/querying/visualizing public or personal project data to build or test hypotheses, discover markers, gain knowledge, and support state-of-the-art breeding applications. These goals and the knowledge required to achieve them represent important pieces of an overall puzzle and therefore require exploration. In summary, harnessing the full potential of pan-genomes for targeted crop breeding is an important goal of the plant research and breeding community. The integration of diverse genomic tools, including pan-genome, marker-assisted selection, genomic selection, and gene editing techniques, will play a pivotal role in developing climate-resilient crops that can adapt to and thrive in the face of climate changes, securing global food security and sustainability (see Figure 5).

5. Outlook, opportunities, and innovations in plant pan-genome research

Plant genomes are often very large and complex, making it difficult to produce high-quality genome assemblies with accurate gene annotations. For over last three decades, genotypic variations have been captured using various genetic markers, including RFLP, RAPD, SNPs, SSRs, microsatellite markers, etc. to establish connections between genotype and phenotype to aid plant breeding and cultivar improvement [168,169,170,171,172,173]. However, the advances in the short-read and long-read-based next-generation sequencing platforms and computational methods required for processing large-scale genomic and transcriptomic data have facilitated the rapid and cost-effective whole genome sequences, including de-novo genome assemblies [7,174,175]. These whole genome sequence comparisons are driving pan-genomic research and, first time in history, allowing researchers to explore intra- and inter-species structural variations at the resolution of nucleotide sequence level. We can expect a significant expansion in the availability of pangenomes across a broader range of plant species. The pangenomes facilitate the identification of genes that are conserved across species and genes that are unique to specific species or a subset of accessions. The availability of species-specific or genus-level pangenomes of crops is crucial for understanding the dynamics of their genome evolution, including the impact of artificial selection and domestication, crop diversification, and adaptation under varied environments [50,174,176,177]. For instance, a super-pangenome of the Citrullus genus comprising of 346 cultivated watermelon accessions and 201 wild accessions suggested that a duplication of the sugar transporter gene ClTST2 was likely selected during domestication for higher fruit sweetness; the wild accessions harbor many genes related to disease resistance compared to cultivated watermelon accessions; and identified domestication sweeps in cultivated watermelon [178].

In addition, plant pan-genome can help to advance the fundamental understanding of the plant kingdom at various scales. It can aid in improving functional annotations of genes and genomes [179,180], and provide insights into specific roles and interactions of different genetic elements. Furthermore, pangenomes could help to understand the evolution of metabolic diversity across diverse taxonomic clades [97]. This knowledge has implications for improving crop traits and developing more resilient and sustainable agricultural practices. Moreover, by analyzing the genomic diversity represented in pangenomes, scientists can understand the distribution and composition of vegetation across different environments. This information is critical for conservation efforts, ecological studies, and developing strategies to protect and sustainably manage plant resources.

Finally, we see a great application of pan-genome data in improving gene annotations and identifying evolutionary conserved sets of genes associated with important agronomic traits. Likewise, comparative genomic studies can help to identify and annotate clade-specific unique genes that determine metabolite compositions of important fruit and vegetable crops or other categories. These specialized metabolic pathways and associated entities can be easily curated in the pathway databases [97,121,123,124,128,181,182]. The workflows for interspecies gene family comparisons, GO annotations, and standard protocols for gene biocuration are very efficient and established; they can easily accommodate the insights gained from intra-species comparisons. For instance, the availability of whole genome sequences of plants has contributed tremendously to the knowledge of gene duplications, gene family evolution, and functional diversification of homologous genes [179,180,183]. Gene Ontology (GO) and Plant Ontology (PO) annotations have played a central role in accessing the potential gene functions [184,185,186]. Furthermore, the comprehensive analysis of plant transcriptomes has helped us to link genes with potential biological processes, pathways, and response to biotic and abiotic stress or stimulants [179,180,187,188,189]. Integrating pangenomes with other omics, such as transcriptomics, epigenomics, proteomics, and phenomics data, will enable a comprehensive understanding of gene regulation and functional mechanisms underlying important agronomic traits. In conclusion, as shown in Figure 5, pan-genomic research is driving functional genomics, evolutionary studies, and biodiversity exploration and consequently, holds great potential for crop improvement, environmental conservation, and sustainable agriculture.

It is important to emphasize that pan-genome construction and whole genome-level comparative analysis require substantial computational infrastructure, expertise in various facets including sequence generation, genome assembly and annotation; and subsequent bioinformatic analyses and data visualization. In general, all these tasks are beyond the capacity of individual research laboratories, and thus, they require extensive infrastructure and bioinformatic support from their institutions and public databases. Public data repositories, databases, genomic resources, and secondary knowledgebases play an essential role in aiding the community of researchers in providing ontologies [184], archiving and annotating genomic data, and supporting analysis and visualization of omics data to support data-driven hypotheses and experimental plans [90,97,115,128]. Here we have reviewed the tools and genomic resources available to the plant research community for conducting pan-genomic research.

We note here that the resources and tools for pan-genome visualization and to support researchers for exploring pan-genomic data are limited and at an early developmental stage. We have witnessed the growing number of publications on crop pan-genomes in the last five years (see Table 2); however, the pan-genomic and genomic diversity data for the majority of the crops are stored and archived with no associated tools/features required for visualization and effective use by other researchers. Thus, advancement and innovations in pan-genomic data visualization, analysis tools, and additional biocuration of genomic data are needed to facilitate meaningful intraspecies and interspecies genomic comparisons. Community biocuration can play an essential role in making sense of the big data and ensuring quality controls at various steps [190]. This will allow researchers to study pan-genomes in more detail and identify genes and genomic regions associated with important traits. Equally important is integrating crop pan-genomes in comprehensive knowledgebases that host analyzed and annotated genomic and pathway data for ensuring the FAIR data policy implementations [111,132].

From a technical aspect, integrating machine learning and artificial intelligence techniques will expedite the analysis of complex pan-genomic datasets, aid in identifying patterns, and accelerate the construction of predictive models and gene biocuration.

Furthermore, the genomic resources are more mature for the model plants and major crop species [132]. But the necessary financial and infrastructure support for minor crops, fruit and vegetable crops, and orphan crops is still insufficient [191,192,193]. Although the cost of sequencing a single genome has decreased significantly in recent years, sequencing multiple genomes can still be prohibitive, a significant barrier to conducting pan-genome research in poorly studied crops or orphan crops.

We expect that this review will help researchers to understand and evaluate the status and strategies being employed for pan-genomic studies in crops, and encourage them to seek necessary collaborations within the community or help to advocate for increased funding for developing the infrastructure, tools, and biocuration of genomic data that is vital for making needed progress. Furthermore, we hope this review can help students and young researchers learn about the status of crop genomic research and the future potential of pan-genomics.