Chromosome-level and high-contiguity genome assembly and annotation of the greasyback shrimp Metapenaeus ensis

Abstract

The greasyback shrimp (Metapenaeus ensis) is an ecologically adaptable and economically important penaeid species. Here, we present its first chromosome-level genome assembly, generated by an integrated approach that combined Oxford Nanopore Technologies (ONT) long-read and Illumina short-read sequencing, followed by Hi-C scaffolding. The long-read data were pivotal for resolving a complex genomic architecture, featuring a genome-wide heterozygosity of 2.46% and a repetitive sequence content of 56.76%. The final assembly spans 1.97 Gb with a contig N50 of 3.84 Mb and a scaffold N50 of 45.37 Mb, 90.07% of the sequences were anchored onto 39 chromosomes. The assembly was annotated with 21,432 protein-coding genes. Assessment with BUSCO using the arthropoda_odb12 lineage dataset indicated high annotation completeness, with 95.6% of core arthropod genes identified as complete, including 89.1% as single-copy. This high-quality genomic resource provides a crucial foundation for advancing evolutionary, ecological, and aquaculture research on this species and related crustaceans.

Background & Summary

The greasyback shrimp (Metapenaeus ensis), commonly known as the sand shrimp or reed shrimp¹, is an economically important penaeid species widely distributed in the Indo-West Pacific, with records from China (particularly the South China Sea), Japan, India, Australia, and other regions within this range. It is characterized by a straight, blade-like rostrum bearing 6–9 dorsal teeth and an unarmed ventral edge. As a generalist feeder, it primarily preys on benthic organisms and inhabits diverse coastal environments—from shallow marine waters to turbid estuaries and mudflats (typically at depths of 8–95 m)^2,3. It shows no strong substrate preference, occupying sandy, muddy, or mixed sediments. Its distinctive morphology, broad ecological adaptability, and aquaculture potential make M. ensis a compelling model for studying crustacean adaptation.

Despite the considerable economic importance of penaeid shrimp in global fisheries, genomic resources for this group remain limited compared with other high-value aquatic species⁴. The genome assembly of penaeid species is often highly fragmented (contig N50 typically < 1 Mb)^5–9, which is largely attributable to their structural complexity, including high repetitive sequence content¹⁰. This inherent challenge is further compounded by the presence of abundant mucopolysaccharides and phosphatases in shrimp tissues, which impede the extraction of high-quality DNA¹¹. These collective challenges have contributed to the current absence of a high-contiguity genome for M. ensis. Consequently, the lack of such a genomic reference has impeded detailed investigations into its population structure, environmental adaptation, and stress resilience. A chromosome-level assembly is therefore a critical prerequisite for advancing a comprehensive biological understanding of this species.

In this study, we employed a hybrid sequencing approach integrating long-read (ONT) and short-read (Illumina) technologies to sequence and assemble the M. ensis genome (Table 1). This draft assembly was subsequently scaffolded to chromosome level using Hi-C data. This produced a high-quality, chromosome-level assembly spanning 1.97 Gb comprising 118 scaffolds with a scaffold N50 of 45.37 Mb, representing a genome of markedly improved continuity. This work demonstrates the substantial advantage of long-read sequencing for resolving complex penaeid shrimp genomes and provides a high-quality foundational resource for future in-depth studies of this species.

Table 1.

Statistical analysis of sequencing data for M. ensis genomes.

Library types	Insert size(bp)	Clean data (Gb)	Read Length (bp)	Coverage (×)
Nanopore Long-read	—	60.67	1,792,244.36(mean)	30.64
Illumina reads	350	130.67	150	66.01
Hi-C reads	350	154.86	150	—
RNA reads	350	70.41	150	—
Total	—	416.61	—	96.65

Methods

Sample collection

A mature female greasyback shrimp (M. ensis) was selected based on the morphology of its external genitalia. Species identification followed the morphological characteristics described in Liu & Zhong (1986), Penaeoid Shrimps of the South China Sea. Key diagnostic features included: (1) the rostrum bearing 6–9 dorsal teeth with a smooth, toothless ventral margin; (2) bright orange-red pleopods; and (3) pereopods with alternating orange-yellow banded intervals. The specimen was collected on November 21, 2024 from the Pearl River Estuary (22.03° N, 113.45° E) at a depth of approximately 15 m on sandy-muddy substrate using a bottom trawl, and was provided by the Provincial Stock Field of the Pearl River Estuary in Zhuhai City, which maintains its broodstock through annual collection from this estuary (Fig. 1a). Muscle tissue was dissected for genomic DNA extraction, while additional tissues (stomach, gill, hepatopancreas, heart, and eye) were collected for RNA extraction and transcriptome sequencing. All samples were immediately flash-frozen in liquid nitrogen and stored at −80 °C to preserve nucleic acid integrity for subsequent Hi-C, whole-genome, and transcriptome library construction.

Fig. 1.

Sample photograph and genome survey analysis of M. ensis. (a) Photograph of the sampled shrimp used for genomic DNA extraction. (b) K-mer frequency distribution plot generated from Illumina short-read sequencing data, used for genome survey analysis including heterozygosity estimation and repeat content prediction.

RNA extraction and sequencing

Total RNA was extracted from stomach, gill, hepatopancreas, heart, and eye tissues using TRIzol reagent. RNA purity was assessed with a NanoDrop One spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), concentration was measured with a Qubit 3.0 fluorometer (Life Technologies, Carlsbad, CA, USA), and integrity was verified by agarose gel electrophoresis. RNA-seq libraries were constructed using the VAHTS Universal V6 RNA-seq Library Prep Kit for MGI (Vazyme, NRM605). Libraries were quantified with Qubit, size-checked on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), and validated by qRT-PCR (effective concentration > 2 nM) prior to paired-end 150 bp sequencing on a DNBSEQ T7 platform (MGI Tech, Shenzhen, China). Raw reads were filtered using fastp (v0.21.0) with the following parameters: adapter trimming, removal of reads containing more than five N bases, and removal of reads in which low-quality bases (Q < 15) accounted for ≥ 40% of the read length, and discarding reads shorter than 30 bp after trimming. After filtering, at least 6 Gb of clean data were obtained per tissue sample for subsequent analysis.

Genome survey and characterization

Genomic DNA was extracted from muscle tissue using an SDS/proteinase K method. Tissue was ground in liquid nitrogen, lysed in SDS lysis buffer containing proteinase K at 50 °C for 1.5 h, and purified by sequential phenol-chloroform extraction and isopropanol precipitation. The DNA pellet was washed with ethanol, dissolved in EB buffer, treated with RNase A, and further purified using a Blood & Cell Culture DNA Mini Kit (Qiagen, 13323). DNA purity was assessed using a NanoDrop One spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), and double-stranded DNA concentration was accurately quantified using a Qubit 3.0 fluorometer (Life Technologies, Carlsbad, CA, USA). Sequencing libraries were prepared from the DNA meeting these quality criteria according to the Illumina TruSeq DNA Sample Preparation Guide (15026486 Rev. C) and sequenced on a NovaSeq X Plus platform with paired-end 150 bp reads, generating 137.75 Gb of raw data. Raw reads were processed using fastp (v0.21.0) for adapter trimming, removal of reads with >5% N bases, low-quality filtering (removing reads with ≥50% of bases having Q ≤ 5). A total of 130.67 Gb of clean data were retained for genome survey analysis.

To assess the fundamental characteristics of the M. ensis genome, we performed a k-mer-based survey analysis using filtered data from short-read sequencing and Jellyfish software (v2.2.10) to generate a 19-mer frequency depth distribution for estimating genome size and heterozygosity¹². The k-mer frequency distribution was profiled with GenomeScope (v2.0)¹³, revealing an estimated haploid genome size of approximately 1.24 Gb. The profile exhibited a prominent heterozygous peak, indicating a high heterozygosity rate of 2.46%. Furthermore, the analysis estimated a high repetitive sequence content of 47.5% (Fig. 1b). These findings delineate a complex genomic architecture for the greasyback shrimp, characterized by a substantial proportion of repetitive elements and high heterozygosity. This complexity necessitates long-read sequencing and a hybrid assembly strategy for resolving heterozygosity and repetitive sequences, ultimately producing a high-quality, contiguous reference genome.

ONT sequencing and assembly

High-molecular-weight (HMW) genomic DNA was extracted from muscle tissue using the standard cetyltrimethylammonium bromide (CTAB) method¹⁴. DNA quality was assessed through a combination of 0.75% agarose gel electrophoresis to confirm fragment size integrity, NanoDrop One spectrophotometry (NanoDrop Technologies, Wilmington, DE) to evaluate purity, and Qubit 3.0 fluorometry (Life Technologies, Carlsbad, USA) with the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Q32854) for accurate concentration measurement, ensuring the extracted DNA met the requirements for long-read sequencing. Sequencing libraries were constructed from 1 μg of HMW gDNA using the Ligation Sequencing Kit (SQK-LSK114) according to the manufacturer’s instructions. The final library was loaded onto a PromethION R10.4.1 flow cell and sequenced on the PromethION sequencer (Oxford Nanopore Technologies, Oxford, UK). Basecalling was performed using Dorado (v5.2.0) with the super-accuracy (SUP) model under default settings, applying a minimum Q-score threshold of 7 for quality filtering. Adapter trimming is automatically enabled in Dorado during basecalling. Following these steps, 60.67 Gb of clean data were generated.

The initial genome assembly and primary error correction were performed using NextDenovo (v2.5.2) with Nanopore long reads under default settings¹⁵, producing a contig-level assembly. This assembly was subsequently polished in two iterative rounds using Racon (v1.4.13) with the same read set under default parameters. Further error correction was conducted by applying Pilon (v1.23) with the -changes - diploid options for two additional rounds, utilizing Illumina short-read data to enhance base accuracy. The Illumina short reads used for polishing achieved a mapping rate of 99.84% and covered 99.06% of the polished assembly. To address the high heterozygosity of the genome, redundant haplotypic contigs were identified and removed using Purge_dups (v1.2.5) with −2 parameter¹⁶, resulting in the purged contig assembly. This purged assembly spans 1.97 Gb and has a contig N50 of 3.84 Mb (Table 2).

Table 2.

Assembly result data statistics.

Metric	ONT initial assembly	Purged contig assembly	Hi-C anchored assembly
Total length(bp)	2,966,431,735	1,975,053,284	1,975,151,684
Number of contigs/scaffolds	2,475 (contigs)	1,102 (contigs)	118 (scaffolds)
GC content (%)	38.491	38.765	38.763
N50 (bp)	3,097,034	3,839,690	45,369,354
N90 (bp)	606,065	921,346	13,817,744
BUSCO completeness (C%)	98.4	97.8	97.8

Mitochondrial genome assembly and phylogenetic analysis

ONT long reads were assembled using Flye (v2.9.2-b1786) with the–nano-raw mode and default parameters¹⁷. Mitochondrial contigs were identified by BLASTN (v2.16.0) using the M. ensis mitochondrial genome (NC_026834.1) as a query¹⁸. Candidate contigs were visualized in Bandage (v0.8.1) to confirm circularization, yielding a complete circular mitochondrial genome¹⁹. Annotation was performed using MitoS2 (v2.1.0) with the invertebrate mitochondrial genetic code (code 5) and Metazoa RefSeq 89 database²⁰. The mitochondrial genome is 15,944 bp in length with a GC content of 34.25%, containing 13 protein-coding genes, 22 tRNA genes and 2 rRNA genes (Fig. 2a).

Fig. 2.

Mitochondrial genome organization and phylogenetic analysis of M. ensis. (a) Circular map of the M. ensis mitochondrial genome. Genes are color-coded by functional category; the inner circle denotes GC content. (b) Bayesian inference phylogenetic tree based on complete mitochondrial genome sequences of 12 species from the family Penaeidae, with Metapenaeopsis dalei (NC_029457.1) as the outgroup. Branch lengths are shown in black and posterior probability values in blue at the corresponding nodes. The tree was constructed using PhyloSuite v2 with MAFFT alignment, Gblocks trimming, and ModelFinder for best-fit substitution model selection, followed by MrBayes analysis (1,000,000 generations) applying the invertebrate mitochondrial genetic code (code 5). GenBank accession numbers for all included species are as follows: Metapenaeus joyneri (NC_042173.1), Metapenaeus affinis (NC_039179.1), M. ensis reference (NC_026834.1), M. ensis from this study (PZ143188.1), Melicertus latisulcatus (NC_040140.1), Litopenaeus vannamei (EF584003.1), Litopenaeus stylirostris (NC_012060.1), Penaeus semisulcatus (MG821354.1), Penaeus monodon (PQ876261.1), Fenneropenaeus chinensis (NC_009679.1), Fenneropenaeus penicillatus (KX891351.1) and Fenneropenaeus indicus (KX462904.1).

Phylogenetic analysis was performed using PhyloSuite v2²¹. Mitochondrial genome sequences of 12 species from the family Penaeidae were retrieved from NCBI GenBank, with Metapenaeopsis dalei designated as the outgroup. Multiple sequence alignment was conducted using MAFFT, followed by trimming with Gblocks. The best-fit nucleotide substitution model was selected using ModelFinder, applying the invertebrate mitochondrial genetic code (code 5) for protein-coding sequences. A Bayesian inference (BI) tree was then reconstructed using MrBayes (1,000,000 generations). All analyses were performed within the PhyloSuite workflow using its integrated tools with default parameters. The resulting phylogenetic tree separated the 12 species into two major clades: one comprising all Metapenaeus species, including our assembled M. ensis mitogenome, and the other containing representatives of Melicertus, Litopenaeus, Penaeus and Fenneropenaeus. The M. ensis mitochondrial genome assembled in this study clustered with the previously published M. ensis reference sequence (NC_026834.1) with a posterior probability of 1.00 and a nucleotide identity of 99.81%, confirming the species identity of the sequenced specimen (Fig. 2b).

Chromosome-level genome assembly

To capture chromatin interaction data, Hi-C libraries were constructed using muscle tissue from the same individual used for nanopore long-read sequencing. The tissue was cross-linked with formaldehyde to preserve the three-dimensional chromatin architecture. The cross-linked DNA was digested with the restriction enzyme DpnII, and the resulting fragments were end-repaired, biotin-labeled with biotin-14-dCTP, and ligated under dilute conditions. Following cross-link reversal, the ligated DNA was purified and sheared to 200–400 bp fragments. DNA ends were phosphorylated, dA-tailed, and ligated to sequencing adapters. Biotinylated fragments were enriched with streptavidin magnetic beads, followed by PCR amplification²². Library quality was assessed using Qubit 3.0 for concentration, Agilent 2100 Bioanalyzer for insert size distribution, and Q-PCR for effective library concentration. The libraries were sequenced on a NovaSeq X Plus platform to generate high-throughput paired-end 150 bp reads.

The raw Hi-C sequencing reads were processed with Fastp (v0.24) for quality control, and the resulting clean data were aligned to the draft genome assembly using HiCUP (v0.8.0)²³. Valid Hi-C interaction pairs were identified using Juicer (v1.6)²⁴, yielding a total of 878,555,968 unique mapped paired-end reads, with a unique mapping ratio of 85.10%. Chromosome-level scaffolding was performed using C-phasing (v0.2.6)²⁵, a comprehensive assembly pipeline for Hi-C chromatin interaction data. Briefly, using the Hi-C read alignments, and C-phasing was used to cluster, order and orient contigs into chromosome-scale pseudomolecules. The assembly was manually reviewed, corrected and visualized using Juicebox (v1.11.08) to address any orientation errors by adjusting contig order and orientation based on Hi-C interaction signals, where stronger interactions reflect closer spatial proximity, guiding the proper placement of contigs²⁶. A total sequence length of 1.78 Gb (90.07% of the assembly) was successfully anchored onto 39 chromosomes (Fig. 3).

Fig. 3.

Hi-C chromatin contact map of the M. ensis genome. The heatmap shows interaction frequencies among the 39 assembled pseudochromosomes (chr1–chr39 along the diagonal). Contact intensity is plotted on a logarithmic scale [Log₁₀(Contact +1)], with warmer colors representing stronger interactions.

Repetitive elements and Non-coding RNA annotation

Repetitive sequences were annotated using a combined strategy of homology-based search and de novo prediction. For homology-based identification, the genome assembly was screened against the RepBase database using RepeatMasker (v4.1.7). Concurrently, a de novo repeat library was constructed using LTR_retriever (v3.0.1) and RepeatModeler (v2.0.6)^27,28. The integrated results revealed that repetitive elements constituted 56.76% of the M. ensis genome. Among these, long terminal repeat (LTR) retrotransposons were the most abundant class, accounting for 26.75% of all repetitive sequences, followed by DNA transposons (23.88%) and long interspersed nuclear elements (LINEs, 18.55%) (Table 3). Genome features were visualized using a Circos plot generated with the circlize package (v0.4.16) in R²⁹. The plot includes chromosome-scale representation, gene density, repeat density and GC content (each calculated in 500-kb windows), together with intra-genomic syntenic blocks identified by MCScanX (v0.8) (Fig. 4)³⁰.

Table 3.

Repeat components in M. ensis genome.

Type	TE Proteins		De novo + repbase		Combined TEs
	Length (bp)	Percentage (%)	Length (bp)	Percentage (%)	Length (bp)	Percentage (%)
DNA	182,747	0.01	471,481,983	23.78	471,585,673	23.88
LINE	44,581,058	2.26	357,568,581	18.10	366,367,820	18.55
SINE	0	0.00	14,018,469	0.71	14,018,469	0.71
LTR	8,164,284	0.41	525,443,709	26.60	528,258,336	26.75
LTR-Gypsy	6,361,810	0.32	70,592,302	3.57	73,256,534	3.71
LTR-Copia	45,516	0.00	3,951,770	0.20	3,993,022	0.20
Other	0	0.00	165,939	0.01	165,939	0.01
Unknown	0	0.00	113,562,509	5.75	113,562,509	5.75
Total	52,926,178	2.68	1,109,940,275	56.20	1,121,149,298	56.76

Fig. 4.

Circos plot depicting genome features of M. ensis. From outer to inner rings: gene density, repeat sequence density, GC content (each calculated in 500 kb non-overlapping windows), and intra-genomic synteny blocks (innermost ring).

The identification of non-coding RNA genes employed specialized tools for each RNA type. Transfer RNA (tRNA) genes were predicted using tRNAscan-SE (v2.0.12) with default parameters³¹. Ribosomal RNA (rRNA) genes were located with RNAmmer (v1.2)³². Additionally, other small non-coding RNAs, including microRNAs (miRNA) and small nuclear RNAs (snRNA), were identified by searching against the Rfam database (v14.10) using the INFERNAL software (v1.1.4) (Table 4)^33,34.

Table 4.

Classification of ncRNAs in M. ensis genome.

Type		Copy	Average length(bp)	Total length(bp)	Percentage (%)
miRNA		134	113	15,132	0.0008
tRNA		3,915	75	292,554	0.0148
rRNA	rRNA	503	305	153,448	0.0078
	18S	12	2,404	28,848	0.0015
	28S	11	6,152	67,668	0.0034
	5.8S	115	118	13,585	0.0007
	5S	365	119	43,347	0.0022
snRNA	snRNA	343	156	53,601	0.0027
	CD-box	148	176	26,042	0.0013
	HACA-box	7	161	1,128	0.0001
	splicing	186	141	26,167	0.0013
	scaRNA	2	132	264	0

Protein-coding gene identification and functional annotation

Protein-coding genes were annotated through an integrated pipeline that combined three complementary approaches. First, transcriptomic-based prediction was performed by aligning filtered RNA-seq reads to the genome using HISAT2 (v2.2.1)³⁵. The resulting alignments were assembled into transcripts by StringTie (v2.2.1)³⁶, followed by coding sequence prediction with TransDecoder (v5.7.0). Second, homology-based prediction was conducted using MiniProt (v0.13) to align protein sequences from four penaeid shrimp species (Fenneropenaeus chinensis, Penaeus japonicus, Penaeus monodon and Penaeus vannamei) against the genome (Fig. 5a)^6,7,9,37,38. Additionally, conserved single-copy ortholog sequences from the BUSCO arthropoda_odb12 lineage dataset were aligned using MiniProt as further homology evidence. Third, ab initio predictions were generated by Augustus v3.5.0 (using models trained from transcriptomic and homology evidence) and by Genscan v1.0 (using its default metazoan model)^39,40. These evidence sources were integrated into a consensus gene set using EvidenceModeler (v2.1.0)⁴¹. In the EVM weights file, weights were assigned per evidence source as follows: TransDecoder (transcript evidence): 20; protein evidence from each of the four penaeid species as independent sources: 20 each; protein evidence from BUSCO: 100; Augustus and Genscan (ab initio predictions): 0.8 each. This weighting strategy was designed to prioritize high-confidence homologs during consensus integration. The consensus set was then subjected to rigorous validation using gffread (v0.12.7) with the -J parameter to filter transcripts lacking canonical start/stop codons or containing in-frame stop codons, and AGAT (agat_sp_fix_cds_phases.pl and agat_sp_filter_gene_by_length.pl) to ensure CDS phase consistency and retain only models with intact open reading frames, yielding a final high-confidence set of 21,432 protein-coding genes for M. ensis (Table 5).

Fig. 5.

Gene structure and functional annotation of the M. ensis genome. (a) Line graph comparing the lengths of gene, CDS, exon, and intron elements between M. ensis and four closely related species. (b) Functional annotation statistics of predicted genes across eight public databases. The bar chart (left) indicates the total number of genes annotated in each database, while the upset plot (right) illustrates the intersection sizes among different annotation sets. (c) Orthologous gene classification across representative penaeid species. The bar chart shows the distribution of single-copy orthologs, multiple-copy orthologs, unique paralogs and other orthologs genes. (d) Synteny analysis between M. ensis and two representative penaeid species (F. chinensis and P. vannamei).

Table 5.

Summary statistics of predicted protein-coding genes in the assembled genome.

Type	Gene Set	Number	Average transcript length (bp)	Average CDS length (bp)	Average exons per gene	Average exon length (bp)	Average intron length (bp)
De novo	Genscan	74,612	14,925.32	1,099.71	4.48	245.52	3,973.77
	AUGUSTUS	136,965	6,631.34	696.53	3.13	222.87	2,792.45
Homolog	F. chinensis	41,742	19,329.77	1,713.47	7.51	228.08	2,705.03
	P. japonicus	45,313	23,173.82	1,950.22	8.59	226.92	2,794.63
	P. monodon	40,756	16,392.15	1,456.87	6.27	232.23	2,832.20
	P. vannamei	32,368	10,623.52	1,132.48	4.09	276.84	3,070.80
RNAseq	TransDecoder	22,800	26,551.06	1,277.43	6.57	521.88	4,151.84
Integration	Maker	28,386	20,714.54	1,359.67	6.45	390.63	3,338.24
Final set*		21,432	26,098.86	1,589.61	7.5	430.95	3,517.39

Functional annotation of the predicted protein-coding genes was performed through a series of sequence similarity and domain database searches. First, homology-based assignments were obtained by aligning all predicted protein sequences against the UniProt and NCBI non-redundant (NR) protein databases using DIAMOND (v2.1.8) with the blastp algorithm⁴². Concurrently, protein domains, families, and conserved sites were identified by searching the InterPro database⁴³, integrating results from its member databases such as Pfam⁴⁴. Finally, to elucidate biological pathways and functional categories, the protein sequences were mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for pathway annotation and to the Gene Ontology (GO) database for functional classification using BLAST+ (v2.11.0). In total, 93.14% of the predicted protein-coding genes were assigned a putative function through annotation by at least one of the above databases (Fig. 5b). To further assess the consistency and completeness of the predicted gene set, orthologous groups were identified using OrthoFinder (v2.5.5) with DIAMOND for sequence similarity searches under default parameters⁴⁵. The results revealed that the M. ensis genome contains 4,033 single-copy orthologs, a count highly consistent with other high-quality Penaeidae genomes such as P. vannamei (4,136) and F. chinensis (4,203). This stability in the core gene set, along with the identification of 2,383 multiple-copy orthologs and 2,813 unique paralogs, demonstrates high annotation completeness and reliability (Fig. 5c).

Data Records

All the raw sequencing data of M. ensis in this study have been uploaded to the NCBI BioProject database under PRJNA1375504⁴⁶. Specifically, the Hi-C, whole-genome short-read, and ONT long-read sequencing data are available in the Sequence Read Archive (SRA) with accession numbers SRR36371134-SRR36371136 and SRR38523471^47–50. The transcriptome sequencing data from five tissues (stomach, gill, hepatopancreas, heart, and eye) are available under BioSample accessions SAMN53693752-SAMN53693756^51–55. The genome assembly has been deposited in GenBank under accession GCA_054960195.1⁵⁶, and the complete mitochondrial genome is available under accession PZ143188.1⁵⁷. The genome assembly, protein sequence and annotation files were deposited at Figshare with doi number https://doi.org/10.6084/m9.figshare.30857777⁵⁸.

Technical Validation

All nucleic acids extracted for this study underwent standard quality control procedures, including assessment by agarose gel electrophoresis, a NanoDrop One spectrophotometer (NanoDrop Technologies, Wilmington, DE), and a Qubit 3.0 Fluorometer (Life Technologies, Carlsbad, CA, USA). The final genome assembly comprises 118 scaffolds with a contig N50 of 3.84 Mb and a scaffold N50 of 45.37 Mb. A total of 1.78 Gb (90.07% of the assembly) was successfully anchored onto 39 pseudochromosomes. Telomeric repeats were identified using tidk (v0.2.41) with the search subcommand targeting the TTAGG repeat motif within 20 kb of each pseudochromosome end⁵⁹. Among the 39 anchored chromosomes, 8 had complete telomeric repeats at both ends (>1,000 copies) and 19 had repeats at a single end (Supplementary Table 1). Assembly accuracy was assessed by realigning Illumina short reads, achieving a high mapping rate (99.84%), sequence coverage (99.06%), and an average depth of 65.27×. RNA-seq reads from five tissues aligned to the assembly with an average mapping rate of 85.46%. The ONT reads had a mean base quality score of Q24.43, and the Illumina short reads had a Q30 proportion of 92.87%, indicating high base-level accuracy for both sequencing platforms. Synteny analysis between M. ensis and two penaeid species (F. chinensis and P. vannamei) revealed clear collinear blocks along the diagonal, supporting the structural accuracy and continuity of our chromosome-level assembly (Fig. 5d).

The completeness and quality of both the genome assembly and its structural annotation were assessed using the Benchmarking Universal Single-Copy Orthologs (BUSCO, v5.4.7) with the arthropoda_odb12 databases⁶⁰. For the genome assembly, 1,631 (97.8%) of the 1,667 conserved arthropod genes were identified as complete, including 90.5% as single-copy and 7.3% as duplicated. For the annotated gene set, 1,593 (95.6%) of the BUSCOs were identified as complete, including 89.1% as single-copy and 6.5% as duplicated. (Table 6). The annotation success rate for the predicted protein-coding genes reached 93.14%, with 19,961 out of 21,432 genes successfully annotated in at least one functional database, including 14,858 (69.33%) assigned to KEGG pathways, 13,345 (62.27%) to Gene Ontology terms, and 18,899 (88.18%) to InterPro domains. The high completeness scores for both the assembly and the annotation demonstrate a highly contiguous assembly and a comprehensive, high-quality gene set.

Table 6.

Statistics of BUSCO assessment in the M. ensis genome.

Type	Assembly		Annotation
	Proteins	Percentage (%)	Proteins	Percentage (%)
Complete BUSCOs	1,631	97.8	1,593	95.6
Complete and single-copy BUSCOs	1,509	90.5	1,485	89.1
Complete and duplicated BUSCOs	122	7.3	108	6.5
Fragmented BUSCOs	18	1.1	30	1.8
Missing BUSCOs	18	1.1	44	2.6
Total BUSCO groups searched	1,667	100	1,667	100

Author contributions

C.Q.H. and P.L. conceived and supervised the study. Y.L., B.M., X.J., S.T.Z., C.H.R., H.Y., J.Y.Y., Y.S.Q. and Y.H.X collected the samples. Y.L. and B.M. performed data analysis and results visualization. Y.L. drafted the manuscript. C.Q.H. and P.L. provided review comments and modification of the manuscript. All authors read and approved the final manuscript.

Funding

This research was supported by the National Key Research and Development Program of China (2023YFD2401701), the Innovative Team Building Project of Guangdong Modern Agricultural Industrial Technology System (2024CXTD25), the Hong Kong Scholars Program (XJWQ2025007) and the Science and Technology Program of Nansha District (NSJL202103).

Code Availability

This study did not rely on novel or custom computational code. All data processing and analyses were conducted using the established bioinformatics software packages detailed in the Methods section, following their standard protocols and manuals (Supplementary Table 2).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41597-026-07627-z.