Skip to main content
NCBI home page
Scientific Data logoLink to Scientific Data
. 2026 Jun 24;13:1406. doi: 10.1038/s41597-026-07697-z

A chromosome-level genome assembly of SAGS Anisodus tanguticus (Maxim.) Pascher (Solanaceae) from the Tibetan region of Sichuan, China

Gang Liu 1,2,✉,#, Dan Su 3,#, Yutong Liu 1,2,#, Zhan Shi 4, Shenzhong Gao 4, Xin He 1,2, Shiyin Zhang 1,2, Heng Deng 4, Jingjing Xiao 3, Yinan Yao 4,✉, Peng Shu 3,✉
PMCID: PMC13655172  PMID: 42342705

Abstract

Anisodus tanguticus (Maxim.) Pascher (A. tanguticus), a cold-tolerant perennial herb in the Solanaceae family, is distributed across China’s Qinghai-Tibet Plateau and extends to Nepal, Bhutan, Sikkim, and India. As a Tibetan medicinal plant, it is used to treat pain, ulcers, etc.; its roots yield antispasmodic and anesthetic compounds, and other parts are used as a feed additive for yaks to enhance cold resistance in northwest Sichuan. In this study, samples were collected from Seda County (northwestern Sichuan, China) for sequencing. Using PacBio HiFi sequencing and Hi-C scaffolding, a high-quality chromosome-scale genome assembly was generated, with a genome size of 1599.64 Mb, a scaffold N50 of 62.01 Mb, and a contig N50 of 38.51 Mb. A total of 24 superscaffolds (93.65% of the genome) were anchored to 24 chromosomes. Compared with previously reported assemblies of A. tanguticus and A. acutangulus, this assembly shows improved scaffold length and completeness. Genome annotation identified 64.95% repetitive elements and 45,930 protein-coding genes, and comparative analysis of four Anisodus genomes revealed conserved patterns of gene density, GC content, LTR, and LINE elements. This study provides the first high-quality chromosome-scale genome resource of A. tanguticus from the Qinghai-Tibet Plateau, supporting studies on phylogeny, genetic diversity, and breeding, as well as further exploration of its genomic basis of high-altitude adaptation.

Background & Summary

Anisodus tanguticus (Maxim.) Pascher (A. tanguticus) is a cold-tolerant perennial herb of the Solanaceae family, distributed across the Qinghai–Tibet Plateau in China (including Tibet, Qinghai, and northwestern Sichuan) and extending to Nepal, Bhutan, Sikkim, and northeastern India. In China, it is a traditional medicinal plant known as ‘Tangchong Nabao’ in Tibetan medicine, historically used to treat pain, ulcers, colitis, gallstones, and trauma. Beyond its traditional medicinal use, the roots of A. tanguticus serve as a crucial raw material for extracting antispasmodic, anesthetic, and bronchodilator compounds, generating significant economic impact. In the northwest Sichuan Plateau region, other parts of the plant are collected and dried by local Tibetan communities for use as feed additives for yaks to protect against winter cold.

Here, we report the first high-quality chromosome-scale genome assembly of A. tanguticus from the Qinghai–Tibet Plateau, with samples collected from Seda County (northwestern Sichuan, China) at elevations above 4,000 m by the Sichuan Academy of Grassland Sciences (SAGS) (Fig. 1). The genome assembly was generated using PacBio HiFi sequencing and chromosome conformation capture (Hi-C) anchoring, resulting in a genome of approximately 1599.64 Mb, with scaffold and contig N50 values of 62.01 Mb and 38.51 Mb, respectively. Contigs were scaffolded into 24 superscaffolds, accounting for 93.65% of the total genome size (Table 1). As illustrated in Fig. 2a, the Hi-C heatmap reveals 24 individual superscaffolds, aligning perfectly with the 24 chromosomes of the SAGS assembly. This SAGS assembly shows improved scaffold length and completeness (99.20% BUSCO) compared with previously published A. tanguticus assemblies and the available A. acutangulus assembly. Genome annotation identified 64.95% repetitive elements and 45,930 protein-coding genes, consistent with other A. tanguticus genomes and supporting the quality of this assembly (Table 1). Moreover, a comparative Circos plot of four Anisodus genomes reveals conserved patterns in gene density, GC content, and LTR/LINE density, highlighting a shared genomic architecture (Fig. 2b). Therefore, this genome resource supports studies on phylogeny, genetic diversity, functional genomics, and breeding, as well as further investigation of the genomic basis of high-altitude adaptation in this polyploid plant.

Fig. 1.

Fig. 1

Open in a new tab

Anisodus tanguticus at the Seda Grassland Ecological Field Scientific Observation and Research Station, Affiliated to Sichuan Academy of Grassland Sciences, Sichuan Province, China. Left: Location of the Grassland Ecological Field Scientific Observation and Research Station (referring to the aforementioned station in Seda, Sichuan). Right: Anisodus tanguticus cultivated at this station.

Table 1.

Comparison of the assemblies of Anisodus tanguticus (SAGS) and the other three published Anisodus genomes.

Category A.tanguticus (SAGS) A.tanguticus (LZU) A.tanguticus (KIB) A.acutangulus (LZU)
Sequencing
Platform PacBio Nanopore Nanopore Nanopore
 Genome-sequencing depth (X) 57.36 105.77 101.86 129.18
Assembly
 Estimated genome size (Mb) 1599.64 1198 1262.53 1260.86
 Assembled genome size (Mb) 1599.64 1249 1230.58 1254.09
 N50 of scaffolds (Mb) 62.01 49.96 51.28 53
 No. of contigs 1886 205 276 209
 N50 of contigs (Mb) 38.51 23.81 25.07 24.64
 GC content of the genome (%) 37.66 37.05 42.37 37
 Anchored to chromosome (%) 93.65 97.47 97.47 99.46
 Complete BUSCOs (%) 99.2 98.1 97.83 94.49
Annotation
 Percentage of repeat sequences (%) 64.95 65.68 66.7 71.79
 LTR rate (%) 50.44 43.07 44.51 57.02
 No. of predicted protein-coding genes 45,930 46,606 44,252 38,211
 Average gene length (bp) 6,250.20 4,846.05 6,868.32 6,151
 Average CDS length (bp) 1,109.08 1,134.68 1155.38 1,243
 Mean exon/intron length (bp) 272.94/1,189.87 216.06/872.93 277.99/1,194.40 210.94/1000.40
 Mean exon number per gene 5.09 5.25 5.22 5.9
Open in a new tab

Fig. 2.

Fig. 2

Open in a new tab

Characteristics of the Anisodus tanguticus Genome. (a) The heatmap represents the 24 chromosomes of the Anisodus tanguticus genome. (b) The genome morphologies and Circos diagram of Anisodus acutangulus and Anisodus tanguticus (from Lanzhou University, LZU), as well as Anisodus tanguticus from the Sichuan Academy of Grassland Sciences (SAGS), Lanzhou University (LZU), and Kunming Institute of Botany, Chinese Academy of Sciences (CAS, KIB). The different tracks (progressing inward) are as follows: (I) gene density; (II) guanine-cytosine (GC) content; (III) density of LTR-Gypsy transposons; and (IV) density of LTR-Copia transposons.

Methods

Plant materials

Fresh young roots, stems, and leaves of Anisodus tanguticus were collected from the medicinal plant cultivation base in Seda County, Sichuan Province, China, a region characterized by a continental highland monsoon climate with an annual mean temperature of −0.16 °C and an average elevation exceeding 4,000 m (Fig. 1), for whole-genome and transcriptome sequencing.

Library construction and sequencing

High-quality genomic DNA was isolated to construct a SMRTbell library with an average insert size of 15-20 kb. This library was sequenced on the PacBio Sequel II platform (Pacific Biosciences, Menlo Park, CA, USA)1, yielding 91.76 Gb of raw HiFi data. Additionally, a Hi-C library was constructed and sequenced with paired-end 150 bp reads (PE150) reads on the Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA), generating 173.54 Gb of raw data. Concurrently, whole-genome short-read sequencing (PE150) and transcriptome sequencing were performed using the DNBSEQ-T7 platform (MGI, Shenzhen, China), which supports high-throughput sequencing with reliable data quality. For whole-genome sequencing, 79.48 Gb of raw data were generated. For transcriptome sequencing, a cDNA library was constructed using the MGI Easy RNA Library Prep Kit with a 350 bp insert size, and paired-end sequencing (PE150 reads) produced 64.93 Gb of raw data (Table 2).

Table 2.

The sequencing overview of Anisodus tanguticus samples.

Data Type Tissue Average coverage (x)
HIFI leaf 57
Hi-C leaf 108
WGS leaf 49
RNA-Seq leaft, root, stem —
Open in a new tab

Genome assembly

For genome preliminary assembly, raw polymerase reads were first subjected to control using the PacBio SMRT Analysis software package (https://www.pacb.com), which removes reads shorter than 50 bp, those with quality values below 0.8, and adapter-contaminated reads. Subsequently, Circular Consensus Sequencing (CCS) reads (HiFi reads) are generated from the filtered data using SMRT Link v9.0 (parameters:–min-passes = 3–min-rq = 0.99), ensuring ≥ 99% base-call accuracy (Phred score ≥ 30). Finally, these HiFi reads were assembled into primary contigs using Hifiasm (v0.16.1; https://github.com/chhylp123/hifiasm)2, which performs all-versus-all read overlap alignment, iterative error correction, and string graph construction to resolve haplotypes and repetitive regions while prioritizing contiguous primary assembly output.

Hi-C assisted scaffolding

Preliminary quality control of Hi-C data was performed using fastp (v0.20.1) with default parameters to remove low-quality reads3, generating high-quality clean reads for downstream analysis. Subsequently, HiCUP (v0.8) was employed to extract valid pairs reads4, representing valid chromatin interaction pairs, which were then used for scaffolding with ALLHiC (v.0.9.14)5. Following ALLHiC scaffolding, large-scale structural variations (e.g., inversions and translocations) were manually corrected through visual inspection of chromatin interaction heatmaps using Juicebox (v2.1543), yielding the final chromosome-level genome assembly6,7. Following the adjustment of chromosome coordinates, the Hi-C contact matrix was visualized using HiC-Pro8 and HiCExplorer (v3.7.2)9 (Fig. 2a).

Sequence consistency assessment

To assess the accuracy of genome assembly, the whole-genome short-read data were aligned to the assembled genome using BWA software10. The alignment rate, genome coverage, and sequencing depth were calculated to evaluate assembly completeness and sequencing uniformity.

Repetitive element annotation

Repeat annotation was performed using a combination of homology-based and de novo approaches. (1) Homology-based identification: Sequences were aligned against the curated RepBase database using RepeatMasker and RepeatProteinMask to detect interspersed repeats homologous to known repetitive elements11. (2) De novo prediction: A genome-specific repeat library was constructed de novo using LTR_FINDER (for LTR retrotransposons)12, Piler13, RepeatScout14, and RepeatModeler. This custom library was subsequently used by RepeatMasker to identify and classify repetitive elements. (3) Tandem repeat detection: Tandem Repeats Finder (TRF) was applied independently to identify simple sequence repeats and tandem arrays15.

In total, 64.95% of the assembled genome was identified as repetitive DNA sequences. The most abundant repetitive elements were retrotransposons, among which long terminal repeat (LTR) retrotransposons accounted for 50.44%, with Copia (28.56%) and Gypsy (16.49%) elements being dominant. The DNA transposons represented 3.67% of the genome (Table 1).

Protein-coding gene prediction

Protein-coding genes were comprehensively annotated through an integrated approach combining de novo prediction, homology-based searches, and transcriptome evidence. EvidenceModeler (v1.1.1)16 was employed to generate a consensus gene set by weighted integration of three annotation sources: (1) Homology-based prediction: utilizing TblastN (v2.2.26)17 for initial alignment and GeneWise (v2.4.1)18 for precise spliced alignment, leveraging cross-species protein homology. (2) De novo prediction: five ab initio tools were executed. Augustus (v3.2.3), Geneid (v1.4), Genescan (v1.0), GlimmerHMM (v3.0.4), and SNAP (v2013.11.29)19–23, each trained on taxon-specific parameters to minimize false positives. (3) Transcriptome-derived evidence: RNA-Seq reads were aligned to the genome using HISAT2 (v2.0.4)24, followed by transcript assembly via Trinity (v2.1.1) and Cufflinks (v.2.2.1)25,26. The final consensus gene set was generated using EvidenceModeler (v1.1.1) by integrating gene predictions from the three evidence sources. The average transcript length and coding sequence length were 5,762.73 and 1,109.07 bp, respectively, with an average of 5.09 exons per gene. The average exon and intron lengths were 225.06 and 1,184.79 bp, respectively. The number of predicted genes in SAGS genome in this study was comparable to those in three previously published genomes, including two Anisodus tanguticus genomes from Lanzhou University (LZU; 46,606 genes) and the Kunming Institute of Botany, Chinese Academy of Sciences (KIB; 44,252 genes), as well as the Anisodus acutangulus genome from Lanzhou University (LZU; 38,211 genes) (Table 1).

Gene function annotation

To assign biological functions, predicted protein-coding genes were aligned against several comprehensive databases, including NCBI non-redundant protein (NR), Swiss-Prot, Kyoto Encyclopedia of Genes and Genomes (KEGG), InterPro, Gene Ontology (GO), and protein families (Pfam)27–31. InterProScan (v5.27.66) was used to search publicly available databases (ProDom, PRINTS, Pfam, SMART, PANTHER, and PROSITE) to annotate protein motifs and domains32–38. A total of 44515 genes (96.90% of the predicted protein-coding genes) were annotated using these databases. Specifically, approximately 90.5%, 65.2%, 65.0%, 94.1%, 57.5% and 69.3% were annotated in NR, Swiss-Prot, KEGG, InterPro, GO, and Pfam, respectively (Table 3).

Table 3.

Functional classification and annotation of mRNA in Anisodus tanguticus.

Database Number Percent (%)
NR 41,554 90.5
Swiss-Prot 29,941 65.2
KEGG 29,846 65
InterPro All 43,199 94.1
Pfam 31,830 69.3
GO 26,397 57.5
Annotated 44,515 96.9
Total 45,930 —
Open in a new tab

Non-coding RNA annotation

Transfer RNA (tRNA) genes were predicted using tRNAscan-SE (v2.0.7) with default parameters, leveraging its enhanced covariance models for high-specificity detection39. Ribosomal RNA (rRNA) sequences were identified via BLASTN searches (E-value threshold < 1e-5) against conserved plant rRNA references, utilizing sequence homology-based approaches. For the annotation of other non-coding RNAs (ncRNAs), including microRNAs (miRNAs) and small nuclear RNAs (snRNAs), Infernal (v1.1.2) was employed to scan the Rfam database (release 14.1)40,41, applying covariance models for structural alignment42. The comprehensive ncRNA annotation pipeline resulted in the prediction of 51,605 ncRNA genes, comprising 451 miRNAs, 12,866 tRNAs, 37,754 rRNAs, and 1,471 snRNAs (Table 4).

Table 4.

Functional classification and annotation of non-coding RNA in Anisodus tanguticus.

Type Copy (w) Average length (bp) Total length (bp) % of genome
miRNA 451 122.1264 55079 0.003443
tRNA 12866 75.50536 971452 0.060729
rRNA rRNA 37754 276.6448 10444449 0.652923
18S 3530 1668.318 5889163 0.368155
28S 12983 142.7766 1853669 0.11588
5.8S 3266 161.2808 526743 0.032929
5S 17975 120.9944 2174874 0.13596
snRNA snRNA 1471 120.8362 177750 0.011112
CD-box 534 104.279 55685 0.003481
HACA-box 101 121.2574 12247 0.000766
splicing 835 131.3665 109691 0.006857
Open in a new tab

Data Records

The raw sequence data and assembled genome sequences generated in this study have been deposited in public repositories. The raw sequence data are publicly accessible at the Genome Sequence Archive (GSA) in the NGDC under accession number CRA02694643, as well as the NCBI Sequence Read Archive (SRA) under accession number SRP69653244. The assembled genome sequences have been deposited in the Genome Warehouse (GWH) under accession number GWHGGGA00000000.145 and in the NCBI GenBank database under accession number JBXYDY01000000046. Additionally, the chromosome length statistics, genome assemble statistics, base content statistics, BUSCO assessment, GEGMA assessment, Illumina reads coverage, single nucleotide polymorphisms, repetitive elements, gene function annotations, ncRNA prediction, gene families, GO term for genes and KEGG term for genes have been deposited at the Figshare database with https://figshare.com/s/12bee600d2489080a20c47.

Technical Validation

The completeness of the final assembled genome was assessed using BUSCO (v4.0.5) software by searching plant-specific databases that contain a total of 1,614 orthologous single-copy genes48,49. The results revealed 99.2% of the BUSCO genes were identified as complete, of which 35.3% were duplicated. Only 0.40% of BUSCO genes were fragmented, and 0.40% were missing from the genome. Following chromosome-scale scaffolding using Hi-C data, we anchored 93.65% of the genome assembly onto 24 chromosomes. These results indicated high genome assembly completeness. CEGMA (v2.0) software employs 248 most highly conserved core eukaryotic genes (CEGs) to assess the extent of comprehensive gene coverage50. The genome assembly showed a high completeness level, with 97.98% (243) of CEGs completely and partially covered, including 95.97% (238) of CEGs considered complete. The filtered short Illumina reads were aligned back to the genome assembly using BWA (v0.7.12-r1039). Approximately 99.66% of the short reads mapped to the genome, and the ratios of heterozygous and homozygous single nucleotide polymorphisms (SNPs) were 0.23% and 3.02e-05%, respectively, indicating that the assembly had high single-base-level accuracy.

Acknowledgements

This project was supported by China Forage and Grass Research System (No. CARS-34).

Author contributions

G.L., D.H. and Y.Y. planned and designed the research; P.S., D.S., Y.L., Z.S., S.G., X.H., S.Z. and J.X. analyzed data; P.S. and G.L. wrote the manuscript.

Data availability

The raw sequencing data and genome assembly of SAGS have been deposited in the National Genomics Data Center (NGDC) under accession number PRJCA041626 (https://ngdc.cncb.ac.cn/bioproject/browse/PRJCA041626), as well as National Center for Biotechnology Information (NCBI) under accession number PRJNA1458118 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1458118).

Code availability

All computational tools and software versions employed in data processing are explicitly documented in the Methods section. For analyses where no custom parameters are specified, default settings were applied throughout.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Gang Liu, Dan Su, Yutong Liu.

Contributor Information

Gang Liu, Email: 350903876@qq.com.

Yinan Yao, Email: yinanyao@swust.edu.cn.

Peng Shu, Email: sp1373751728@tmmu.edu.cn.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

  1. NCBI Sequence Read Archivehttps://identifiers.org/ncbi/insdc.sra:SRP696532 (2026).
  2. Shu, P. A chromosome-level genome assembly of SAGS (Anisodus tanguticus (Maxim.) Pascher). Figshare 10.6084/m9.figshare.30628253 (2025). [DOI] [PMC free article] [PubMed]

Data Availability Statement

The raw sequencing data and genome assembly of SAGS have been deposited in the National Genomics Data Center (NGDC) under accession number PRJCA041626 (https://ngdc.cncb.ac.cn/bioproject/browse/PRJCA041626), as well as National Center for Biotechnology Information (NCBI) under accession number PRJNA1458118 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1458118).

All computational tools and software versions employed in data processing are explicitly documented in the Methods section. For analyses where no custom parameters are specified, default settings were applied throughout.

Articles from Scientific Data are provided here courtesy of Nature Publishing Group

RESOURCES