Background & Summary

Agastache rugosa, a perennial plant belonging to the Lamiaceae family, is widely distributed in Korea, China, Taiwan, and Japan. In Korean traditional medicine, the aerial part of A. rugosa, known as “Gwakyang”, is prescribed for various ailments, such as miasma, cholera, anorexia, and vomiting1. A. rugosa produces phenolic compounds such as rosmarinic acid, which has antioxidant and antibacterial properties2,3,4,5. In addition to its uses in traditional herbal medicine, A. rugosa leaves are used as a spice or vegetable and its flowers as a tea ingredient6. Desta et al. assessed the antioxidant activity of various parts of A. rugosa—including the flowers, leaves, stems, and roots—and found that the leaves, flowers, and roots exhibited notably strong antioxidant properties7.

Previous research on A. rugosa has primarily concentrated on its secondary metabolites3,4, phenylpropanoid-biosynthetic genes8,9,10, and cell culture11,12. To date, there are no whole genome sequences available for A. rugosa, and only transcriptome data have been published13. An integrated analysis of its metabolites and genome will provide insight into chemotype breeding of A. rugosa and improve its economic value in the market.

In this study, we assembled the chromosome-level genome of A. rugosa using Nanopore sequencing and Hi-C technology. The final genome assembly had a scaffold N50 of 52.15 Mbp, totaling 410.67 Mbp. With integration of Hi-C data, nine pseudochromosomes were generated, accounting for 89.1% of the entire predicted genome. The first chromosome-scale genome of A. rugosa provides a foundational genetic resource for breeding programs targeting enhanced production of secondary metabolites like rosmarinic acid and essential oils. This genome assembly bolsters the efficiency of genotyping methods such as GBS, facilitating more precise QTL analysis or GWAS, which are crucial for optimizing agricultural traits.

Methods

Sampling and sequencing

A breeding line, AG34, of A. rugosa, sourced from a specific population in the field, was chosen for reference genome sequencing and assembly. This line was derived from original natural accessions obtained from the Chungbuk National University (Korea). Young leaf samples were collected once during the vegetative stage after being grown in a greenhouse for three months. Leaf tissue samples were stored at −80 °C and used for DNA extraction, whole genome sequencing, and Hi-C library construction. DNA was extracted using the Biomedic Plant gDNA extraction kit (#BM20211222A, Korea) following the manufacturer’s instructions.

An Oxford Nanopore Technology (ONT) sequencing library was constructed using the ONT genomic ligation sequencing kit SQK-LSK110 (ONT, UK). ONT sequencing was performed using the flow cell vR9.4 (FLO-MIN106) and GridION platform operated with MinKNOW Core 4.4.3 following the manufacturer’s instructions. We obtained 55.9 Gb of raw genomic data. Guppy v5.0.17, embedded in MinKNOW14, was used to convert raw ONT sequencing data (FAST5 files) to FASTQ format using the default parameters of the high-accuracy method. All ONT sequencing procedures were conducted by Phyzen Co. (www.phyzen.com, Korea). Paired-end (PE) Illumina sequencing was also conducted with the NovaSeq6000 platform after constructing a standard Illumina paired-end library. We obtained 115.5 Gb of raw data from Illumina sequencing.

Total RNA was extracted from leaf tissue of the same material used for the genome sequencing of A. rugosa, and the transcriptome was sequenced on the Illumina NovaSeq6000 platform by Macrogen Co. (www.macrogen.com, Korea). The RNA reads were used for gene annotation.

Sequence trimming and genome size estimation

ONT data were trimmed using Porechop (v.0.2.3, https://github.com/rrwick/Porechop) with default parameters to remove adaptors and chimeric sequences. Raw Illumina sequencing data were trimmed using fastp (v.0.21.0, https://github.com/OpenGene/fastp) with default parameters. The amount of trimmed Illumina PE sequencing data was 97 Gb, which was used for further genome size estimation based on k-mer analysis. An optimal k-mer value of 19 was calculated by Jellyfish (v2.0)15, and the genome size was estimated using GenomeScope (v2.0)16. The estimated genome size of A. rugosa based on k-mer analysis was 460.89 Mbp, which is slightly smaller than the 520 Mb previously reported using flow cytometry17. The heterozygous rate was 0.55%, and the repeat rate was 62.21% (Fig. 1).

Fig. 1
figure 1

The result of K-mer analysis. (a) 19-mer frequency distribution in A. rugosa genome. The X-axis is the k-mer depth, and Y-axis represents the frequency of the k-mer for a given coverage. (b) Statistics of K-mer analysis.

Full size image

Contig assembly

The first round of de novo assembly was performed using NextDenovo assembler (v.2.3.1, https://github.com/Nextomics/NextDenovo) with default parameters, employing only preprocessed 55,923,595,489 bp of ONT data(~121X of estimated genome size, 460Mbp). Assembled contigs were then polished using NextPolish (v1.3.1, https://github.com/Nextomics/NextPolish) with trimmed Illumina PE sequencing data. Haplotigs were removed using Purge Haplotigs18 with default parameters. The assembly statistics improved, with fewer contigs and increased minimum, average contig lengths, and N90 (see Table S1). Finally, a draft genome assembly was generated with 221 contigs totaling 410.65 Mbp, with a contig N50 of 3.85 Mbp (Table 1).

Table 1 Assembly statistics of A. rugosa.
Full size table

Chromosome-level genome assembly using Hi-C data

A Hi-C library of A. rugosa was constructed for chromosome assembly using the ProximoTM Hi-C Plant Kit (Phase Genomics, United States) following the manufacturer’s instructions. A total of 30.77 Gbp of clean Hi-C data were generated and aligned to the assembled contigs using BWA-MEM (v0.7.17)19 with -5SP and -t 8 options specified. Chromosome-level scaffolding was performed with the Phase Genomics Proximo Hi-C genome scaffolding platform based on the LACHESIS method20, and sequences were anchored to nine pseudochromosomes with chromosome lengths ranging from 27.7 Mb to 73.6 Mb. Our chromosome-scale assembly coincides with that from a previous karyotype analysis, as the base chromosome number of Agastache species is reported to be nine, and A. rugosa is a diploid species21,22. Additional manual correction of the chromatin contact matrix was performed using Juicebox (https://github.com/aidenlab/Juicebox). The nine pseudochromosomes were clearly identified by distinct interaction signals in the Hi-C interaction heatmap (Fig. 2), and the final assembled genome was 410.68 Mbp, with a scaffold N50 of 52.15 Mb, accounting for 89.1% of the predicted genome size based on the k-mer analysis (Table 1 and Fig. 3). The assembled genome sizes of Lamiaceae species show a wide range of variation: A. rugosa in this study (410.68 Mbp), Perilla frutescens var. hirtella (676.94 Mbp)23, P. frutescens var. frutescens (1.2 Gbp)23, Salvia hispanica (321.47 Mbp)24, and Salvia splendens (805.9 Mbp)25.

Fig. 2
figure 2

Hi-C contact map the chromosome-level assembly of A. rugosa. The intensity of interactions was calculated using a bin size of 140 K.

Full size image
Fig. 3
figure 3

Overview of genome features of the A. rugosa. Syntenic block among inter-chromosome were analyzed with MCScanX. (a) Gene distribution, (b) Repeat percentage(%), (c) Gypsy (red line) and Copia (blue line) LTR distribution (%), (d) GC content(%).

Full size image

Assessment of the genome assemblies

The completeness of the assembled genome was evaluated using BWA-MEM (v0.7.17)19 and Benchmarking Universal Single-Copy Orthologs (BUSCO, v5.2.1)26 with the embryophyta_odb10 lineage dataset. Approximately, 98.04% of the Illumina short read were aligned to genome, of which 89.6% of reads were properly mapped. The BUSCO analysis showed that the assembled draft genome sequence contained 1,596 (98.9%) complete BUSCOs, including 1,533 (95.0%) single-copy BUSCOs, 63 (3.9%) duplicated BUSCOs, and 7 (0.4%) fragmented BUSCOs (Table 2).

Table 2 Result of the BUSCO assessment of A. rugosa.
Full size table

Repeat annotation

The de novo repeat families were identified with RepeatModeler27, and by LTR_retriever28, then repetitive sequences were masked using RepeatMasker 4.0.9 (http://www.repeatmasker.org). A total of 561,061 repeat elements were identified, accounting for 61.65% of the A. rugosa genome. Among the various repeat elements, Copia and Gypsy, which are long terminal repeats (LTRs), were dominant in the genome, accounting for 14.98% and 13.91%, respectively (Table 3).

Table 3 Repetitive elements annotation in A. rugosa.
Full size table

Gene prediction and annotation

Gene prediction involved a combination of evidence-based annotation methods and ab initio prediction using repeat-masked assembly sequences. RNA-Seq data were assembled by Trinity and used for the transcript set. Additionally, protein data from four related Lamiaceae species were obtained from the NCBI. The first round of gene prediction was performed using MAKER (v3.01.03)29 with evidence data, the transcript set and the protein data from the four related species. The ab initio gene predictions were conducted on only the first gene models with sufficient evidence (AED of 0.25 or less) using GeneMark-ES (v4.38)30, SNAP (v2006-07-28)31, and Augustus (v3.3.2)32. Final gene predictions were confirmed again based on the first gene model and ab initio gene model using MAKER3 (v3.01.03)29 and EvidenceModeler (v1.1.1)33. In total, 26,430 protein-coding genes were predicted and annotated, with an average gene length of 1,184 bp (Table 4). The complete BUSCOs of predicted gene set were calculated as 98.9%.

Table 4 Summary of gene annotation.
Full size table

The predicted genes of A. rugosa were functionally annotated by comparing their similarities against those in the NCBI nonredundant (nr) protein database and the reference genome Araport11 of Arabidopsis thaliana using DIAMOND (v0.9.30.131)34 with an E-value cutoff of 1E-5. Conserved protein domains were predicted by InterProScan (v5.34-73.0)35. Gene Ontology analysis was conducted using the Blast2GO command line (v.1.4.4), and genes were assigned to metabolic pathways by comparing them to those in the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database36 using the KEGG Automatic Annotation Server (KAAS) webtools (v2.1)37. A total of 24,624 genes were successfully annotated for A. rugosa, accounting for 93.2% of all predicted genes (Table 4 and Fig. 4). Predicted gene models were comparable to four other Lamiaceae species in aspects such as gene count, average CDS length, average exons per gene, and average exon and intron length (Table 5).

Fig. 4
figure 4

Venn diagram of the number of genes from A. rugosa with homology or functional classification using multiple public databases.

Full size image
Table 5 The comparison of the gene models annotated from A. rugosa genome and other Lamiaceae.
Full size table

Ortholog and phylogenetic analysis

Orthologs between A. rugosa and eight other plants (seven from the order Lamiales: S. hispanica24, Salvia miltiorrhiza38, P. frutescens var. hirtella23, Paulownia fortune39, Erythranthe guttata40, Andrographis paniculata41, and Genlisea aurea42, along with one outgroup, Vitis vinifera43) were identified using OrthoFinder (v2.5.4)44. The sequences for these plants were sourced from the NCBI database (http://www.ncbi.nlm.nih.gov/). From these, 371 single-copy orthologous genes were extracted, concatenated, and aligned using the Multiple Alignment program for amino acid or nucleotide sequences (MAFFT)45. We then constructed a maximum likelihood phylogenetic tree of these orthologous genes using RAxML (v8.2.12)46 under the JTT model, Gamma Distributed With Invariant Sites (G + I), with a bootstrap value of 1000. Four species, namely A. rugosa, S. hispanica, S. miltiorrhiza, and P. frutescens var. hirtella, all of which belong to the Lamiaceae family, clustered in the same clade. Notably, A. rugosa exhibited a closer relation to the two Salvia species (Fig. 5). These findings are consistent with previous phylogenetic studies based on the chloroplast genome47.

Fig. 5
figure 5

Phylogenetic relationship of Lamiales species.

Full size image

Data Records

The genomic Illumina sequencing data were deposited in the Sequence Read Archive at the NCBI (SRR24282004)48.

The genomic Nanopore sequencing data were deposited in the Sequence Read Archive at the NCBI (SRR24282001)49.

The transcriptome Illumina sequencing data were deposited in the Sequence Read Archive at the NCBI (SRR24282003)50.

The Hi-C sequencing data were deposited in the Sequence Read Archive at the NCBI (SRR24282002)51.

The final chromosome assembly was deposited in GenBank at the NCBI (GCA_031470985.1)52.

The annotation result of gene structure, functional prediction, and final chromosome assembly were deposited in the Figshare database (https://doi.org/10.6084/m9.figshare.22730084)53.

Technical Validation

The integrity and concentration of the extracted DNA and RNA were assessed with a TapeStation 2200 and an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA), respectively. In a comparative context, the complete BUSCO value for A. rugosa (98.9%) exceeds those of P. frutescens var. frutescens (92.7%)23, P. frutescens var. hirtella (92.5%)23, S. splendens (92.0%)25, and S. hispanica (97.8%)24, underscoring its relative completeness and quality within the Lamiaceae family.