Chromosome-level genome assembly of the wild Chinese yam Dioscorea opposita Authors: Tiegang Yang, Huijun Wang, Peng Wang, Yulong Zhao, Xiangyang Li, Guixia Shi, Tiejun Guo, Jie Tang, Guixiao La Published: 2026 ABSTRACT Dioscorea opposita, a perennial dioecious vine endemic to East Asia, holds substantial economic value owing to its dual role as a source of both food and traditional medicine. In this study, we constructed a high-quality, chromosome-level reference genome for this species by integrating PacBio, BGI, and Hi-C sequencing technologies. The assembled haploid genome is approximately 430.73 Mb, 94.33% of the sequences were anchored onto 20 pseudochromosomes. Our analysis confirmed that this species not only underwent a recent hexaploidization event (2n = 6x = 120), but also carried genomic signatures originating from earlier polyploidization events. We annotated 24,960 protein-coding genes, of which 93.58% were functionally annotated. Repetitive sequences accounted for 65.93% of the genome, with LTR elements being the most abundant component (48.00%). BUSCO confirmed high completeness of both the genome assembly and gene annotation. This genomic resource provides a crucial reference for studying sex determination, adaptive evolution, and genomic diversity within Dioscorea and lays a solid foundation for genetic improvement, germplasm conservation, and sustainable utilization of this key plant. Background & Summary Chinese yam (Dioscorea opposita Thunb.) belongs to the genus Dioscorea L. in the family Dioscoreaceae. The genus Dioscorea comprises approximately 630 perennial, dioecious vine species, and its members have emerged as one of the most important tuber crops worldwide due to their starch- and saponin-rich tubers1,2. In China, D. opposita is also renowned for its dual role as a food and medicinal resource3 and demonstrates a broad spectrum of physiological and pharmacological activities, including immunomodulatory, antitumor, hypoglycemic, antioxidant, and anti-aging effects3–5. Notably, it has a history of cultivation and medicinal use spanning centuries. This long-standing history has given rise to numerous populations within the species. Among these, the population known as “Huai yam”—which originated from Shennong Mountain in Jiaozuo, China—is widely regarded as the most nutritious and medicinally valuable, with an annual yield exceeding 260,000 tons. However, in recent years, the local yam germplasm has become increasingly admixed with introduced yam varieties, compromising both the quality of the produce and the ability to morphologically distinguish it from the original “Huai yam”. To address this issue, we collected genotype WT42 from wild yam population in Shennong Mountain and performed chromosome-level genome assembly and annotation. This work aims to provide robust genomic support for the conservation and breeding of this valuable genetic resource. To date, genome-level assemblies have been completed for several Dioscorea species, including D. alata6,7, D. rotundata8, D. dumetorum9, D. nipponica10, D. zingiberensis11,12, and D. tokoro13. These genomic resources have greatly advanced our understanding of genetic diversity, phylogeny, adaptive evolution, diosgenin biosynthesis, anthracnose resistance genes, and sex determination mechanisms across the genus. Nevertheless, genomic investigations of D. opposita remain scarce, which limits comprehensive insights into its genetic background, evolutionary history, and molecular mechanisms underlying sex determination. In this study, we constructed a high-quality chromosome-level reference genome for Dioscorea opposita by integrating PacBio HiFi long reads (56.83 × coverage), BGI short reads (692.99 ×), Hi-C data (339.73 ×), and full-length transcriptome sequencing (Table 1). The assembled genome has a total size of approximately 430.73 Mb with a scaffold N50 of 19.82 Mb, and 94.33% of the sequences were anchored onto 20 pseudochromosomes (Tables 2, 3). We annotated 24,960 protein-coding genes, and repetitive sequences accounted for approximately 65.93% of the genome (Tables 4, 5). As the first published high-quality reference genome for D. opposita, this work provides an essential genomic foundation for exploring sex determination, genome evolution, and adaptive divergence within the Dioscorea genus. Moreover, it establishes a robust scientific basis for future research on genetic improvement, resource conservation, and sustainable utilization of Dioscorea species.Table 1Summary of sequencing data used for D. opposita genome assembly.Read typeRead base (Gb)Number of reads/spotsMean read length (bp)Depth (x)Q20 (%)Q30 (%)BGI genomic reads325.901,086,330,479PE150692.9997.8993.12HiFi genomic reads37.172,006,09918,53056.8395.2288.77Hi-C reads146.35487,825,274PE150339.7397.1792.73Table 2Key features of the D. opposita genome assembly and annotation.StatisticsValueAssembled genome size (bp)430,723,595Sequence assigned to chromosomes (bp)406,311,163Anchored rate (%)94.33Number of scaffolds311Scaffold N50 (bp)19,823,099Number of contigs350Contig N50 (bp)8,165,153Number of chromosomes20Number of gap-free chromosomes3Number of gaps39Protein-coding genes24,960Table 3Summary of the 20 pseudochromosomes in the D. opposita genome.ChromosomeLength (bp)GC (%)Number of contigsQV scores (BGI reads)QV scores (HiFi reads)Chr0119,822,09938.18351.9065.94Chr0218,861,78936.89150.8163.78Chr0318,317,47937.59448.7061.93Chr0425,032,65237.61353.3062.92Chr0527,695,25736.86251.4061.59Chr0618,175,80437.32550.7861.42Chr0726,021,30438.24353.0265.90Chr0820,511,33436.73350.5661.56Chr0914,942,79236.65152.1463.49Chr1018,389,52837.61351.8566.09Chr1118,248,93137.58352.3161.33Chr1217,634,61037.97552.3662.80Chr1318,321,43338.06251.0760.46Chr1422,205,95237.55552.3462.24Chr1523,411,92638.07352.4460.82Chr1615,920,79137.68251.5763.32Chr1721,178,56637.79250.1962.99Chr1822,540,68738.05452.4062.70Chr1926,527,49937.45153.8262.99Chr2012,550,73039.41452.4164.23Total406,311,16337.645946.7560.28Table 4Annotation and distribution of repetitive sequences in the D. opposita genome.TypeCountLength (bp)Percentage (%)DNA195,30439,041,9199.06LINE84,83825,459,1375.91SINE45251,0980.01LTR431,935206,737,24548.00 LTR/Gypsy259,817152,030,53735.30 LTR/Copia126,78943,430,72010.08 LTR/Others45,32918,987,0944.41Tandem repeats267,5672,879,7120.67Others89,60522,308,2055.18Total1,069,701283,958,03665.93Table 5Statistics of gene function annotation for the D. opposita genome.NumberPercentage (%)Protein-coding genes24,960100Annotated23,38293.68 InterPro17,91371.77 GO17,06368.36 KEGG22,39989.74 Pathway8,83235.38 Swissprot16,85767.54 TrEMBL22,50190.15 NR23,35893.58Unannotated1,5786.32 Table 1Summary of sequencing data used for D. opposita genome assembly.Read typeRead base (Gb)Number of reads/spotsMean read length (bp)Depth (x)Q20 (%)Q30 (%)BGI genomic reads325.901,086,330,479PE150692.9997.8993.12HiFi genomic reads37.172,006,09918,53056.8395.2288.77Hi-C reads146.35487,825,274PE150339.7397.1792.73 Summary of sequencing data used for D. opposita genome assembly. Table 2Key features of the D. opposita genome assembly and annotation.StatisticsValueAssembled genome size (bp)430,723,595Sequence assigned to chromosomes (bp)406,311,163Anchored rate (%)94.33Number of scaffolds311Scaffold N50 (bp)19,823,099Number of contigs350Contig N50 (bp)8,165,153Number of chromosomes20Number of gap-free chromosomes3Number of gaps39Protein-coding genes24,960 Key features of the D. opposita genome assembly and annotation. Table 3Summary of the 20 pseudochromosomes in the D. opposita genome.ChromosomeLength (bp)GC (%)Number of contigsQV scores (BGI reads)QV scores (HiFi reads)Chr0119,822,09938.18351.9065.94Chr0218,861,78936.89150.8163.78Chr0318,317,47937.59448.7061.93Chr0425,032,65237.61353.3062.92Chr0527,695,25736.86251.4061.59Chr0618,175,80437.32550.7861.42Chr0726,021,30438.24353.0265.90Chr0820,511,33436.73350.5661.56Chr0914,942,79236.65152.1463.49Chr1018,389,52837.61351.8566.09Chr1118,248,93137.58352.3161.33Chr1217,634,61037.97552.3662.80Chr1318,321,43338.06251.0760.46Chr1422,205,95237.55552.3462.24Chr1523,411,92638.07352.4460.82Chr1615,920,79137.68251.5763.32Chr1721,178,56637.79250.1962.99Chr1822,540,68738.05452.4062.70Chr1926,527,49937.45153.8262.99Chr2012,550,73039.41452.4164.23Total406,311,16337.645946.7560.28 Summary of the 20 pseudochromosomes in the D. opposita genome. Table 4Annotation and distribution of repetitive sequences in the D. opposita genome.TypeCountLength (bp)Percentage (%)DNA195,30439,041,9199.06LINE84,83825,459,1375.91SINE45251,0980.01LTR431,935206,737,24548.00 LTR/Gypsy259,817152,030,53735.30 LTR/Copia126,78943,430,72010.08 LTR/Others45,32918,987,0944.41Tandem repeats267,5672,879,7120.67Others89,60522,308,2055.18Total1,069,701283,958,03665.93 Annotation and distribution of repetitive sequences in the D. opposita genome. Table 5Statistics of gene function annotation for the D. opposita genome.NumberPercentage (%)Protein-coding genes24,960100Annotated23,38293.68 InterPro17,91371.77 GO17,06368.36 KEGG22,39989.74 Pathway8,83235.38 Swissprot16,85767.54 TrEMBL22,50190.15 NR23,35893.58Unannotated1,5786.32 Statistics of gene function annotation for the D. opposita genome. Methods Plant materials, sample collection, and sequencing A wild male Dioscorea opposita (designated WT42) was sampled at Shennong Mountain, Jiaozuo, Henan Province, China (35°13′30′′ N, 112°48′10′′ E). The fresh young leaves were collected, immediately flash-frozen in liquid nitrogen on-site, and subsequently stored at −80 °C. The germplasm was also transplanted to the Institute of Chinese Herbal Medicines, Henan Academy of Agricultural Sciences for further study (Fig. 1).Fig. 1Morphological characteristics of Dioscorea opposita. (A) Entire plant; (B) Mature tubers; (C) Bulbils, also referred to as “Lingyuzi”, which serve as propagation materials for the subsequent growing season. Morphological characteristics of Dioscorea opposita. (A) Entire plant; (B) Mature tubers; (C) Bulbils, also referred to as “Lingyuzi”, which serve as propagation materials for the subsequent growing season. High-molecular-weight genomic DNA was isolated using the DNAsecure Plant Kit (DP320; Tiangen Biotech, China). The purity and concentration of the DNA were assessed using a NanoDropTM 2000 spectrophotometer (Thermo Fisher Scientific, USA) and a Qubit® 3.0 Fluorometer (Thermo Fisher Scientific, USA), respectively. DNA integrity was evaluated by 0.8% agarose gel electrophoresis. A short-read sequencing library with an insert size of 350 bp was constructed using the VAHTS Universal DNA Library Prep Kit for MGI (NDM607; Vazyme Biotech, China) according to the manufacturer’s instructions. The library was sequenced on the DNBSEQ-T7 platform (Beijing Genomics Institute, China), generating approximately 325.90 Gbp of paired-end 150 bp reads (Table 1). For long-read sequencing, genomic DNA was sheared into fragments of 15–20 kb. A HiFi library was then prepared using the SMRTbell Express Template Preparation Kit 3.0 (Pacific Biosciences, USA) and subjected to circular consensus sequencing (CCS) on the PacBio Revio platform, yielding 37.17 Gbp of HiFi data with an average read length of 18.53 kb (Table 1). To achieve a chromosome-level assembly, we employed high-throughput chromosome conformation capture (Hi-C) technology. The Hi-C library was constructed following the protocol described previously14. Briefly, fresh leaf tissue was fixed with 1% formaldehyde for 10 minutes to prevent over-crosslinking. Subsequently, the fixed tissue was digested using the restriction enzyme MboI, followed by ligation and purification steps. Throughout the entire experimental procedure, harsh vortexing, high-speed centrifugation, and repeated freeze-thaw cycles were strictly avoided to preserve DNA integrity. Additionally, agarose gel electrophoresis was performed after both formaldehyde crosslinking and enzymatic digestion-ligation steps to monitor DNA quality at these critical stages. The purified DNA was sheared into fragments of 300–500 bp, and sequencing was conducted on the DNBSEQ-T7 platform. This generated approximately 146.35 Gbp of paired-end 150 bp Hi-C data (Table 1). For transcriptome sequencing, total RNA was isolated from leaf, stem, flower, and tuber tissues using the RNAprep Pure Plant Kit (Tiangen Biotech, China). Full-length cDNA was synthesized with the SMARTer PCR cDNA Synthesis Kit (Takara Biotech, China), followed by the preparation of SMRTbell libraries using the Pacific Biosciences DNA Template Prep Kit 3.0 (Pacific Biosciences, USA). Full-length transcriptome sequencing was carried out on the PacBio Revio platform, resulting in 351,248 transcripts with an average length of 1,236 bp. Karyotype analysis and genome size estimation Due to the long history of cultivation and extensive breeding of Dioscorea opposita, inconsistencies have been reported regarding its ploidy and chromosome numbers15–17. To clarify this, metaphase chromosome spreads were prepared from root tips following the method described by Zhou et al.15. Specifically, root tips (1.0–1.5 cm in length) were excised and treated with 500 µL of 0.002 mol/L 8-hydroxyquinoline solution in the dark for 1 hour. Samples were then fixed in Carnoy’s fluid for 24 hours. After fixation, root tips were rinsed with 1 mL of ddH₂O for 10 minutes and digested at 37 °C for 18 minutes in an enzyme mixture containing 3% cellulase and 0.1% pectolyase. Post-digestion, the tissues were washed again with 1 mL ddH₂O for 15 minutes and preserved in Carnoy’s solution. Chromosome slides were prepared via the squash technique. Finally, the chromosomes were stained with 5% Giemsa solution and examined under a microscope (Olympus, Tokyo, Japan). The somatic chromosome number was unambiguously determined to be 2n = 120 (Fig. 2A).Fig. 2Karyotype analysis and genome size estimation of D. opposita. (A) Chromosome morphology (2n = 120); (B) Flow cytometry analysis. Maize B73 (2C DNA = 4.22 Gb) was used as an internal reference. All measurements were conducted with three biological replicates. The mean fluorescence intensities of maize B73 and yam WT42 were 27.67 (27.51, 27.03, 28.47) and 15.22 (16.22, 14.26, 15.18), respectively, resulting in a mean fluorescence ratio of 0.55 (0.59, 0.53, 0.53). Based on these data, the 2 C genome size of D. opposita was estimated to be 2.32 ± 0.14 Gb. (C) Genome survey of D. opposita via k-mer analysis (k = 21, 31, 41, 51) using GenomeScope2. This analysis revealed that this wild yam is hexaploid, with an estimated haploid genome size of approximately 351.83 Mb and a heterozygosity rate of 5.58%. Karyotype analysis and genome size estimation of D. opposita. (A) Chromosome morphology (2n = 120); (B) Flow cytometry analysis. Maize B73 (2C DNA = 4.22 Gb) was used as an internal reference. All measurements were conducted with three biological replicates. The mean fluorescence intensities of maize B73 and yam WT42 were 27.67 (27.51, 27.03, 28.47) and 15.22 (16.22, 14.26, 15.18), respectively, resulting in a mean fluorescence ratio of 0.55 (0.59, 0.53, 0.53). Based on these data, the 2 C genome size of D. opposita was estimated to be 2.32 ± 0.14 Gb. (C) Genome survey of D. opposita via k-mer analysis (k = 21, 31, 41, 51) using GenomeScope2. This analysis revealed that this wild yam is hexaploid, with an estimated haploid genome size of approximately 351.83 Mb and a heterozygosity rate of 5.58%. The genome size of the sequenced individual was determined using flow cytometry, following the method established by Dai et al.18 Briefly, fresh leaf tissue of D. opposita was co-chopped with that of maize (B73; 2C = 4.22 Gb)19, which served as an internal reference standard, and stained with propidium iodide. The experiment was performed in triplicate, yielding a mean fluorescence ratio (D. opposita/maize) of 0.55. Based on this ratio, the 2C genome size of D. opposita WT42 was calculated to be 2.32 Gb (4.22 × 0.55; Fig. 2B). To further characterize the genome, a genome survey was conducted using k-mer analysis of high-quality short reads. Raw reads were initially processed with fastp v1.0.120 under the following parameters: '–qualified_quality_phred 20–unqualified_percent_limit 40–n_base_limit 5–length_required 50–detect_adapter_for_pe'. K-mer frequency distributions for k = 21, 31, 41, and 51 were calculated using Jellyfish v2.2.1021, and genome characteristics were estimated using GenomeScope v2.022. GenomeScope v2.0 was specifically optimized for complex genomes, particularly those with high heterozygosity and polyploidy. K-mer analysis indicated that this wild yam is hexaploid, with an estimated haploid genome size of approximately 351.83 Mb and a heterozygosity rate of 5.58% (Fig. 2C). Collectively, these results demonstrate that the wild yam is a hexaploid (2n = 6x = 120) with a haploid genome size of approximately 350–400 Mb. Chromosome-level assembly An primary assembly was constructed from PacBio high-fidelity long reads using Hifiasm v0.2423 under default parameters. To enhance the accuracy and completeness of this initial assembly, quality-filtered BGI short reads were mapped to the assembly and subjected to two rounds of error correction using Pilon v1.2424. To achieve chromosome-level assembly, the Hi-C scaffolding approach was utilized. The Hi-C data yielded 146.35 Gb of sequence, providing 339.73× genome coverage (Table 1). The Juicer v1.6.225 pipeline was employed to align Hi-C reads to the initial assembly, remove duplicates, and generate a contact matrix quantifying interaction frequencies across genomic regions. Subsequently, 3D-DNA26 was executed with default parameters to scaffold the assembly. Manual inspection and adjustment were then performed in Juicebox v2.6.225 to ensure accuracy and integrity at the chromosome scale (Fig. 3). Based on collinearity analysis with the published D. alata7 genome, the identity of each pseudochromosome in this wild yam assembly was determined, and the chromosomes were designated as Chr01 through Chr20 (Fig. 7).Fig. 3Hi-C contact map of the D. opposita chromosome-level genome assembly. The heatmap illustrates pairwise interaction frequencies between genomic regions derived from Hi-C data used for chromosome-level scaffolding. The color intensity corresponds to the strength of the genomic interaction frequency (darker colors indicate higher interaction frequencies). Hi-C contact map of the D. opposita chromosome-level genome assembly. The heatmap illustrates pairwise interaction frequencies between genomic regions derived from Hi-C data used for chromosome-level scaffolding. The color intensity corresponds to the strength of the genomic interaction frequency (darker colors indicate higher interaction frequencies). The final assembly has a total length of 430.72 Mb, with a contig N50 of 8.17 Mb and a scaffold N50 of 19.82 Mb. It contains 39 gaps and exhibits a GC content of 37.64%. A total of 406.31 Mb (94.33%) of the sequence was anchored onto 20 pseudochromosomes (Tables 2, 3). Integrating these metrics with independent lines of evidence, we conclude that this yam is a hexaploid species. The conclusion is supported by the cytological count of 2n = 120 chromosomes (Fig. 2A), the Hi-C interaction map which clearly resolved 20 pseudochromosomes (Fig. 3), and a genome size consistent across both flow cytometry (Fig. 2B) and genome survey analyses (Fig. 2C). Furthermore, the highly homologous and nearly indistinguishable nature of the subgenomes suggests that it is an autohexaploid. This high-quality genome assembly provides a robust foundation for subsequent gene annotation and functional genomic studies (Fig. 4).Fig. 4Chromosomal features of D. opposita. The 20 chromosomes are ordered by size (megabases) and displayed in 500 kb intervals. Tracks include: (A) chromosome ideogram; (B) gene density; (C) transposable elements (TEs); (D) Copia elements; (E) Gypsy elements; (F) GC content; (G) collinearity blocks within the genome. Chromosomal features of D. opposita. The 20 chromosomes are ordered by size (megabases) and displayed in 500 kb intervals. Tracks include: (A) chromosome ideogram; (B) gene density; (C) transposable elements (TEs); (D) Copia elements; (E) Gypsy elements; (F) GC content; (G) collinearity blocks within the genome. Repetitive sequences annotation Repetitive sequences were identified using a hybrid approach that combined de novo prediction and homology-based searches. For de novo prediction, RepeatModeler v2.0.627 was employed, which integrated multiple specialized tools to construct a species-specific repetitive sequence library. LTR retrotransposons represent the most prevalent class of repetitive elements in plant genomes, accounting for over half of the genomic content in complex species such as millet28 and wheat29. To accurately identify these elements, a non-redundant LTR library was constructed using LTR_FINDER v1.0.730 and LTR_retriever v2.9.031. Tandem repeats were annotated using Tandem Repeat Finder v4.0932. For the homology-based approach, RepeatMasker v4.1.5 (https://www.repeatmasker.org/) was used to search against the RepBase database. Additionally, TE-related proteins were predicted using RepeatProteinMask v4.1.5. Finally, all repetitive sequences identified through the aforementioned methods were merged and clustered with CD-HIT v4.8.133 to generate a non-redundant repetitive sequence library. In total, repetitive sequences spanned 283.96 Mb, accounting for 65.93% of the D. opposita genome. Among these, LTR retrotransposons constituted the majority (48.00%), while DNA transposons, long interspersed nuclear elements (LINEs), and short interspersed nuclear elements (SINEs) represented 9.06%, 5.91%, and 0.01% of the genome, respectively (Table 4; Fig. 4). Gene annotation and functional prediction Protein-coding genes in D. opposita were predicted using an integrated approach that combined transcriptomic evidence, homology-based searches, and de novo prediction. First, full-length PacBio HiFi transcripts derived from leaf, stem, flower, and tuber tissues were used to predict open reading frames (ORFs) with PASA v2.5.334. For homology-based annotation, protein sequences from three related species—D. alata7, D. cayenensis subsp. rotundata8, and D. zingiberensis12—were aligned to the D. opposita genome using BLAST v2.13.0+35, followed by processing with Exonerate v2.4.036 under the “protein2genome” model. De novo gene prediction was performed on the repeat-masked genome using Augustus v3.5.037 and GlimmerHMM v3.0.438. The gene sets generated by these three methods were integrated and consolidated using MAKER v3.01.0339. The final gene models were further refined with PASA to annotate untranslated regions (UTRs) and identify alternative splicing events. In total, 24,960 protein-coding genes were predicted, with an average gene length of 4,529 bp and 5.11 exons per gene. These genomic characteristics were consistent with those of closely related species. For comprehensive functional annotation, the predicted protein sequences were subjected to similarity searches against multiple databases. DIAMOND v2.1.840 was used to query the InterPro41, SwissProt42, TrEMBL42, NR, KEGG43, and GO44 databases under stringent parameters (--max-target-seqs 10; --ultra-sensitive-evalue 1e-5). Approximately 93.68% of the predicted genes (23,382 genes) were functionally annotated in at least one database, providing a robust foundation for subsequent functional genomic studies (Table 5). Noncoding RNA annotation Noncoding RNAs (ncRNAs) were identified using a combination of specialized tools. Specifically, transfer RNAs (tRNAs) and ribosomal RNAs (rRNAs) were identified with tRNAscan-SE v2.0.1245 and RNAmmer v1.246, respectively. For additional ncRNA types, including small nuclear RNAs (snRNAs) and microRNAs (miRNAs), prediction was performed by searching the Rfam database47 using Infernal v1.1.448. In total, 2,014 rRNAs, 95 snRNAs, 68 miRNAs, and 353 tRNAs were annotated in the D. opposita genome (Table 6).Table 6Statistics of non-coding RNA (ncRNA) annotation for the D. opposita genome.TypeCopyTotal length (bp)Percentage (%)rRNA2,0145,597,7211.30 28S8773,953,8390.92 18S8871,615,1990.37 5S25028,6830.01snRNA9511,9250.00 CD-box525,5170.00 HACA-box57170.00 splicing385,6910.00 scaRNA000.00miRNA688,5170.00tRNA35326,6620.01 Table 6Statistics of non-coding RNA (ncRNA) annotation for the D. opposita genome.TypeCopyTotal length (bp)Percentage (%)rRNA2,0145,597,7211.30 28S8773,953,8390.92 18S8871,615,1990.37 5S25028,6830.01snRNA9511,9250.00 CD-box525,5170.00 HACA-box57170.00 splicing385,6910.00 scaRNA000.00miRNA688,5170.00tRNA35326,6620.01 Statistics of non-coding RNA (ncRNA) annotation for the D. opposita genome. Phylogenetic analysis To elucidate the evolutionary history of this wild yam, we selected 13 additional representative species with key phylogenetic positions. Using OrthoFinder v3.1.249 with an e-value cutoff of 1e-5, we identified 392 single-copy orthologous protein clusters. These protein sequences were aligned with MAFFT v7.50850 under the parameters '-genafpair -maxiterate 1000'. For species tree reconstruction, ASTRAL v5.15.551 was employed to coalesce individual maximum-likelihood gene trees across 100 iterations. Divergence times were estimated in BEAST v2.6.352, calibrated with two fossil-based time constraints: the crown age of angiosperms was set to 200 Mya, and the divergence between monocots and eudicots was set to 160 Mya53,54. The well-supported phylogenetic tree shows that Dioscorea split into two clades following a divergence event around K-Pg boundary (65 Mya). Among these, the studied wild yam is most closely related to D. alata, having diverged from it around 19 Mya (Fig. 5A).Fig. 5Phylogenetic analysis of Dioscorea and representative species. (A) Phylogenetic tree and estimated divergence times. Numbers at nodes represent bootstrap values (values of 100 are omitted). (B) Distribution of synonymous substitution rates (Ks). Peaks within a species indicate the timing of whole-genome duplication events, while peaks between species correspond to speciation times. Phylogenetic analysis of Dioscorea and representative species. (A) Phylogenetic tree and estimated divergence times. Numbers at nodes represent bootstrap values (values of 100 are omitted). (B) Distribution of synonymous substitution rates (Ks). Peaks within a species indicate the timing of whole-genome duplication events, while peaks between species correspond to speciation times. To detect whole-genome duplication (WGD) events, we performed a synonymous substitution (Ks)-based analysis. Initially, Ks values for all collinear gene pairs were calculated with KaKs_calculator v2.0. Then, duplication signals were visualized and assessed by constructing block-based Ks dot plots with the WGDI v0.1.6 package55. The Ks distribution revealed that all Dioscorea species underwent two rounds of diploidization. These events occurred in the most recent common ancestor (MRCA) of the Dioscorea genus, subsequent to its divergence from Acanthochlamys but prior to the radiation within Dioscorea (Fig. 5B). Data Records All raw whole-genome sequencing data have been deposited in the National Genomics Data Center (NGDC)56 under the BioProject accession number PRJCA047769. Specifically, the genomic PacBio sequencing data, genomic BGI sequencing data, Hi-C sequencing data, and RNA sequencing data were deposited in the Genome Sequence Archive (GSA) of NGDC with the respective accession numbers: CRA03112257, CRA03111658, CRA03112459, and CRA03112560. The genome assembly has been deposited in the NCBI GenBank under the GenBank accession JBTTGV00000000061. The genome assembly and annotation also have been deposited at FigShare with the doi number 10.6084/m9.figshare.3038455362. Technical Validation Multiple approaches were employed to evaluate the quality and accuracy of the D. opposita genome assembly. The completeness of the assembly was assessed using BUSCO v6.0.062 with the embryophyte_odb12 dataset (containing 2,026 single-copy orthologs) in “genome” mode. The analysis revealed that 1,884 (93.0%) BUSCOs were complete, including 1,811 single-copy and 73 duplicated orthologs; only 35 (1.7%) were fragmented, and 107 (5.3%) were missing. When assessed in “protein” mode, the completeness was slightly lower: 1,753 (86.5%) BUSCOs were complete (1,697 single-copy and 56 duplicated), with 119 (5.9%) fragmented and 154 (7.6%) missing orthologs (Fig. 6).Fig. 6Assessment of genome assembly and annotation completeness for D. opposita using BUSCO. Evaluation was performed against the embryophyte_odb12 dataset, which contains 2,026 single-copy orthologs. Assessment of genome assembly and annotation completeness for D. opposita using BUSCO. Evaluation was performed against the embryophyte_odb12 dataset, which contains 2,026 single-copy orthologs. To verify assembly accuracy and completeness, both BGI short reads and PacBio HiFi long reads were mapped back to the assembly. Using BWA-MEM v2.363, 96.73% of the BGI clean reads aligned successfully, achieving an average depth of 692.99× and 99.89% coverage as determined by bamdst v1.1.0. With Minimap2 v2.30-r128764, the mapping rate for PacBio HiFi reads reached 99.68%, with an average depth of 56.83× and 99.99% coverage (Table 1). These high mapping rates confirm the high accuracy and completeness of the genome assembly. Base-level consensus accuracy was further evaluated using Merqury v1.365, which leveraged 19-mers from both BGI short reads and HiFi long reads. Consensus quality values (QV) were calculated for each chromosome individually. QV scores derived from BGI short reads ranged from 60.46 to 66.09 across all 20 chromosomes, while those from HiFi long reads ranged from 48.70 to 53.82 (Table 3), indicating high base-level accuracy in the final assembly. To validate the large-scale structural accuracy of the assembly, collinearity analysis was conducted between the D. opposita assembly and the published genomes of four related species: D. alata7, D. rotundata8, D. zingiberensis12, and D. tokoro13. This analysis confirmed the identity of each pseudochromosome (Fig. 7). Furthermore, the observed 1:4 syntenic relationship across these genomes provides direct evidence that all analyzed species within this genus have undergone two rounds of diploidization66.Fig. 7Genomic synteny analysis between D. opposita and other congeneric species. Genomic synteny analysis between D. opposita and other congeneric species.