A highly contiguous chromosome-level genome assembly of the Chilean jack mackerel Trachurus murphyi

Abstract

The Chilean jack mackerel (Trachurus murphyi) is an economically important pelagic fish widely distributed across the South Pacific Ocean. Due to intense fishing pressure and the lack of high-quality genomic resources, the development of sustainable fisheries management strategies has been hindered. Here, we present a highly contiguous chromosome-level genome assembly of T. murphyi using PacBio and Hi-C sequencing data. The assembled genome spans 818.31 Mb, with a contig N50 length of 24.89 Mb and a scaffold N50 length of 34.34 Mb. The assembly is anchored on 24 pseudo-chromosomes, covering 95.80% of the genome. BUSCO analysis indicates a high level of completeness at 99.0%. Telomeric repeat motifs were detected at 12 single chromosome ends, whereas putative centromeric regions were identified on 12 chromosomes. Repetitive sequences account for 23.86% of the total genome, and 47,734 protein-coding genes were annotated, 90.98% of them with assigned function. This high-quality genome assembly serves as a valuable resource for studying the genetic diversity and population structure of the Chilean jack mackerel and provides a solid foundation for future management and conservation efforts.

Background & Summary

Jack mackerels are pelagic fish species of the genus Trachurus, found in all world´s oceans except the Arctic and Antarctic waters. They belong to the Carangidae family, a group of ray-finned fishes that includes jacks, pompanos, and trevallies¹, and are classified into 14 species². These species play a crucial role in marine ecosystems due to their trophic position in the marine food web³ and their importance as valuable fishery resources worldwide⁴. They are commercially exploited in many countries and are consumed fresh, frozen, smoked, or canned. Among Trachurus species, the Chilean jack mackerel (T. murphyi) fishery has been one of the largest globally, with catches peaking at nearly 5 million tons in the mid-1990s⁴. T. murphyi is distributed across the South Pacific Ocean⁵, allowing it to exploit a wide range of environmental conditions due to its presence in regions with highly heterogenous oceanographic and prey dynamics^6–8. T. murphyi is valued by consumers for providing an excellent source of high-quality proteins, polyunsaturated fatty acids, vitamins, and minerals⁹. It could be that for those reasons, this species has been harvested for human consumption since at least the Early Holocene^10,11.

Over the past decades, the Chilean jack mackerel population has declined significantly due to a combination of factors, including overfishing and variability with the El Niño-Southern Oscillation (ENSO)^4,12–15. Consequently, the implementation of effective fisheries management strategies is crucial. Previous studies have explored the species´ fishery biology and genetics^16–20. Population genetic analyses have revealed that T. murphyi constitutes a single, large panmictic population across its entire geographic range^18,21–24. Additionally, recent research has identified a ZW sex determination system in T. murphyi²⁵. Despite these findings, genomic and transcriptomic resources for T. murphyi remain limited to the mitochondrial genome, which was being only recently described²⁶. Whole genome sequence assemblies are currently available only for the two Trachurus species, the Atlantic jack mackerel (Trachurus trachurus)²⁷ and the Japanese jack mackerel (Trachurus japonicus)²⁸ but no homologous assembly is available for T. murphyi.

Despite its ecological and economic importance, the lack of a high-quality, chromosome-level reference genome has limited detailed investigation of population structure, adaptive variation, and genomic responses to environmental change in Trachurus murphyi. Therefore, a contiguous genome assembly is essential to support fisheries management, conservation strategies, and comparative studies within Carangidae. Here, we have generated a nearly complete and highly contiguous chromosome-level assembly of the T. murphyi genome using PacBio Hi-Fi long-read and Omni-C/Hi-C sequencing. The generation of such a resource provides a foundational framework for future molecular and evolutionary research.

Methods

Sample preparation and DNA sequencing

The specimen used in this study was provided by commercial fisheries as a fish carcass. It was not subjected to experimental manipulation and handling when it was alive in the context of this work. According to Chilean regulations, institutional animal ethics committee approval is not required for the use of fishery-derived specimens. A single adult individual of Trachurus murphyi (fork length of 45.4 cm and weights 1.395 kg) was obtained from commercial fisheries in the Biobío Region, Chile (36.94°S, 73.75°E) in June 2022 and used for genome sequencing. The specimen exhibited no visible signs of diseases, parasitic infection, or developmental abnormalities, and showed normal external morphology (Fig. 1A). Once collected, the individual was transported to the laboratory on ice to minimize DNA degradation and muscle tissue was dissected and immediately stored at -80 °C until shipped on dry ice to Cantata Bio (https://cantatabio.com/) for DNA extraction, genomic DNA sequencing, and Hi-C library construction. The sex of the individual could not be determined through visual gonad inspection prior to DNA extraction and sequencing. No molecular sex-identification assays were performed. Given this limitation, the assembled genome reported here represents a chromosome-scale baseline pending sex-aware phasing.

Fig. 1.

Chromosome-level genome assembly of Trachurus murphyi. (A) Adult T. murphyi specimen used for whole-genome sequencing. (B) Square-binned blob plot showing contig distribution according to GC proportion (x-axis) and sequencing coverage (y-axis, log10 scale). Each square represents contigs with a defined bin, colored according to taxonomic assignment against the NCBI nr database, with blue indicating similarity to Chordata sequences and gray indicating no significant similarity, and scaled by cumulative span. Marginal histograms display the distribution of GC content and coverage across contigs. The inset highlights the main density of contigs to improve visualization and interpretability. The predominance of contigs assigned to Chordata, together with the absence of non-chordata clusters, supports the lack of detectable contamination in the assembly. (C) Omni-C/Hi-C interaction heatmap displaying genomic coordinates along both axes in megabases (Mb). Color intensity reflects contact frequency between genomic loci. Strong interaction signals along the diagonal indicate high intra-chromosomal contact frequency and support chromosome-scale scaffolding. Chromosome boundaries indicated by grid lines. (D) BUSCO assessment of the T. murphyi genome using the Actinopterygii database (actinopterygii_odb10; n = 3,640 orthologs). Categories represent Complete (C), including Single-copy (S) and Duplicated (D), Fragmented (F), and Missing (M) orthologs. Exact BUSCO values are provided in the main text for clarity.

High-quality DNA, in terms of quantity and purity, was used to construct two PacBio Hi-Fi SMRTbell^® libraries and one Dovetail Omni-C^® sequencing library for assembling the T. murphyi genome. The PacBio libraries were sequenced with circular consensus sequencing (CCS) mode on a PacBio Sequel II system, generating 7,058,275 High-Fidelity (HiFi) reads, totaling 86.5 Gb of data, which covered the genome at 108-fold, with a N50 read length of 12,194 bp. For the Dovetail Omni-C^® library, chromatin was first crosslinked with formaldehyde within the nucleus before extraction. The fixed chromatin was then digested with DNAseI, followed by end repair and ligation to a biotinylated bridge adapter, enabling proximity ligation of adapter-containing ends. After ligation, crosslinks were reversed, and the DNA was purified. Biotin that was not internal to ligated fragments were removed, and the sequencing library was prepared using NEBNext Ultra enzymes and Illumina-compatible adapters. Biotin-containing fragments were enriched using streptavidin beads before PCR amplification. The final library was sequenced on an Illumina HiSeqX platform using 2 × 150 bp reads, generating 87,721,376 read pairs, amounting to 28.6 Gb of data, with a coverage of 32.1-fold across the genome.

Chromosome-level genome assembly with PacBio and Hi-C data

The genome assembly was generated using PacBio HiFi reads and Hifisiam²⁹ (v0.15.4-r347) with default parameters to generate both the primary and haplotigs-alternative sequences for regions where heterozygosity was detectable within primary contigs using long reads (Table 1). Given the high per-base accuracy of HiFi reads and the consensus accuracy achieved by hifisiam^29,30, no additional polishing steps were applied. The assembled contigs were then screened for potential contamination using the BlobToolKit³¹ (v4.4.3) and the BLAST non-redundant database (Fig. 1B). Contigs identified as potential contamination were removed from the assembly (Table 1). Finally, purge_dups³² (v1.2.5) was applied to eliminate haplotigs and overlapping contigs. Following these steps, the contig-level genome assembly spans 802.27 Mb, comprising 337 contigs with an N50 length of 24.89 Mb (Table 1).

Table 1.

Summary statistics of the assembled contigs and scaffolds of Trachurus murphyi.

Features	Hifiasm assembly	Hifiasm + BlopToolKit assembly	Purge_dups assembly	Dovetail HiRise assemly
Total sequences	562	562	337	523
Total bases (bp)	818,306,126	818,306,126	802,744,728	818,265,206
Longest sequence (bp)	50,717,993	50,717,993	50,717,993	43,616,551
N50 length (bp)	24,885,125	24,885,125	24,885,125	34,335,234
N90 length (bp)	6,328,993	6,328,993	6,633,185	25,125,839
Ns (%)	0.00	0.00	0.00	0.00

To achieve a chromosome-level assembly, we used the contig-level genome assembly and Dovetail OmniC^® reads as input for HiRise, a software pipeline specifically designed to scaffold genome assemblies using proximity ligation data³³. Briefly, Dovetail OmniC® reads were mapped to the contig-level genome assembly using BWA (https://github.com/lh3/bwa) with a minimum quality threshold of 50 (MQ > 50). HiRise then analyzed the separation of Dovetail OmniC^® read pairs mapped within draft scaffolds to generate a likelihood model for genomic distances. This model was used to identify and correct putative misjoins, score prospective joins, and execute those joins above a set threshold. Following these steps, we generated a Hi-C contact map and manually reviewed it using Juicebox³⁴ (v1.11.08) to correct any misjoins and position unplaced contigs.

No gap-filling or telomere extension procedures were applied during genome assembly. The differences in total assembly length between the contig-level assembly (802.27 Mb) and the final scaffolded assembly (818.31 Mb) reflects differences in the assembly stage accounting rather than sequence addition. Specifically, the contig-level assembly length corresponds to the primary assembly after haplotig removal using purge_dups³², whereas the final assembly includes the 24 pseudo-chromosome and all retained scaffolds following Hi-C-based scaffolding, including unplaced sequences that were excluded from intermediate statistics (Fig. 1C, Table 1, Table 2). Consistent with this, the final assembly contains no ambiguous bases (Ns), indicating no artificial gap filling was performed (Table 1).

Table 2.

Chromosome lengths and telomere statistics of Trachurus murphyi.

Chromosome name	Length (bp)	GC (%)	Telomere	Left length (bp) / Right length (bp) of telomere sequence
Chr1	43,616,551	42.17	No	0 / 0
Chr2	40,697,839	41.62	No	0 / 0
Chr3	39,753,857	41.81	Left	3,420 / 0
Chr4	39,403,342	41.91	Right	0 / 4,326
Chr5	38,686,146	42.20	No	0 / 0
Chr6	36,848,658	42.03	Right	0 / 1,782
Chr7	36,663,105	41.79	No	0 / 0
Chr8	36,082,001	42.16	Right	0 / 1,362
Chr9	35,151,152	42.16	No	0 / 0
Chr10	34,807,739	42.01	Left	1,212 / 0
Chr11	34,335,234	41.94	No	0 / 0
Chr12	33,135,089	42.20	Left	5,460 / 0
Chr13	32,997,167	41.80	No	0 / 0
Chr14	32,784,410	41.70	No	0 / 0
Chr15	32,188,078	42.45	No	0 / 0
Chr16	30,992,529	42.29	Right	0 / 3,660
Chr17	30,849,155	41.81	No	0 / 0
Chr18	30,703,906	41.91	Left	1,266 / 0
Chr19	28,653,432	42.02	Left	828 / 0
Chr20	27,799,838	42.19	Right	0 / 1,104
Chr21	25,177,033	42.28	Right	0 / 5,094
Chr22	25,125,839	42.19	Right	0 / 5,760
Chr23	24,528,291	42.13	No	0 / 0
Chr24	12,959,388	43.22	No	0 / 0
Unplaced scaffolds	34,308,866	48.44	-	-
ChrMT	16,561	46.59	-	-

Identification of telomere and centromere regions in the chromosome-level genome assembly

Given the high quality and gapless nature of our chromosome-level genome assembly (Table 1), we sought to determine whether it meets the criteria for a telomere-to-telomere (T2T) genome assembly. Telomeric repeat identification was performed by sequence motif detection only, and no sequence extension or telomere reconstruction was attempted. To identify telomeric regions, we searched the Chilean jack mackerel genome for the (TTAGGG)n sequence, the canonical telomere repeat found in most animals³⁵. Telomere identification was carried out using the TeloExplorer function of the quarTeT³⁶ (v1.2.5) software with the “-c animal” option. A visual representation of the telomere distribution is provided in Fig. 2A.

Fig. 2.

Telomeric and centromeric regions in the chromosome-level genome assembly of Trachurus murphyi. (A) Chromosome ideograms scaled according to chromosome length (Mb). Telomeric repeat regions (blue circles) and putative centromeric regions (red rectangles) were identified using the TeloExplorer and CentroMiner modules implemented in quarTeT³⁶. Telomeric signals were detected at single chromosome ends where present, whereas centromeric regions were identified on 12 chromosomes. (B) Omni-C/Hi-C chromatin interaction maps at 50 kb resolution for representative chromosomes. Axes represent genomic coordinates in kilobases (kb). Putative centromeric regions are characterized by reduced interaction frequency and distinct patterns consistent with centromere-associated chromatin organization.

To identify centromeric regions, which are characterized by large stretches of repetitive satellite DNA or tandem repeats ranging in size from hundreds of kilobases to megabases³⁷, we used the CentroMiner tool of the quarTeT³⁶ (v1.2.5) software with default parameters. Briefly, this tool detects all tandem repeat monomers in a genome assembly and selects those most likely to be centromeric repeats on each chromosome based on their period and copy number. These monomers are then clustered to reduce redundancy and aligned to the corresponding chromosomes. Continuous matching regions are added to the candidates set, and scores are assigned based on total match length and retrotransposon content to ultimately determine the locations and predominant monomer repeats³⁶. Our analyses revealed that putative centromeric regions range in length from 0.3 Mb to 2.1 Mb (Fig. 2A, Table 3). These regions are characterized by a low density of protein-coding genes and a high abundance of repetitive elements, with LTR/Gypsy, LINE/L2 and simple repeats representing the predominant repeat classes (Table 3), consistent with the typical structural arragement of centromeres³⁸. In addition, centromeric regions display large interaction-depleted (“blank”) domains in Omni-C contact maps (Fig. 2B), a hallmark feature of centromeres that has also been reported in other gap-free or chromosome-level genome assemblies where centromeres have been resolved^39,40. No single conserved core centromeric repeat unit was identified; instead, centromeres appear to consist of heterogeneous arrays of dispersed repeats, reflecting the species-specific and rapidly evolving nature of centromeric DNA in teleost fishes^41–43. Therefore, centromere positions were inferred at regional level based on repeat composition and Hi-C interaction patterns, rather than by a single consensus repeat sequence. The inferred centromere locations are indicated on chromosome ideograms (Fig. 2A).

Table 3.

Summary statistics and presence of repetitive sequences in the putative centromere regions of T. murphyi.

Chromosome	Start position (bp)	End position (bp)	Length (Mb)	LTR/Gypsy (%)	LINE/L2 (%)	Simple repeats (%)	# of genes
Chr1	29,500,000	31,600,000	2.1	1.73	1.24	0.0	112
Chr2	—	—	—	—	—	—	—
Chr3	10,900,000	12,200,000	1.3	2.57	6.93	0.0	63
Chr4	9,100,000	9,600,000	0.5	4.22	0.03	0.11	16
Chr5	34,100,000	34,600,000	0.5	1.79	0.65	0.43	19
Chr6	29,900,000	31,200,000	1.3	3.30	1.30	0.0	104
Chr7	—	—	—	—	—	—	—
Chr8	5,900,000	6,500,000	0.6	3.81	0.51	0.38	26
Chr9	—	—	—	—	—	—	—
Chr10	—	—	—	—	—	—	—
Chr11	—	—	—	—	—	—	—
Chr12	—	—	—	—	—	—	—
Chr13	—	—	—	—	—	—	—
Chr14	21,100,000	21,600,000	0.5	1.14	0.27	0.0	9
Chr15	24,700,000	25,100,000	0.4	0.92	0.19	0.68	12
Chr16	7,400,000	8,000,000	0.6	0.62	0.22	1.24	26
Chr17	7,200,000	7,600,000	0.4	0.97	0.24	0.81	23
Chr18	—	—	—	—	—	—	—
Chr19	—	—	—	—	—	—	—
Chr20	9,000,000	9,500,000	0.5	1.65	0.30	0.22	28
Chr21	7,900,000	8,200,000	0.3	1.76	0.03	1.22	8
Chr22	—	—	—	—	—	—	—
Chr23	—	—	—	—	—	—	—
Chr24	—	—	—	—	—	—	—

Telomeric repeat sequences were identified at one end of 12 out of 24 pseudo-chromosomes, while no chromosome exhibited telomeric repeats at both ends. Centromeric regions were identified on multiple chromosomes based on repeat composition and Omni-C/Hi-C interaction patterns.

Repetitive sequence annotation

To identify repetitive sequences in the Chilean jack mackerel genome, we generated a custom repeat library using a combination of de novo and homology-based approaches. This included RepeatModeler⁴⁴ (v2.0.1), TransposonPSI⁴⁵ (v1.0.0), LTRharvest⁴⁶ (v1.6.2), LTRdigest⁴⁷ (v1.6.2), TIRvish⁴⁸ (v1.6.2), MITE Tracker⁴⁹ (v1.0), and HelitronScanner⁵⁰ (v1.1), all executed with default parameters. The individual repeat libraries were concatenated, and sequences sharing more than 80% similarity were merged to remove redundancy using USEARCH⁵¹ (v11.0.667). The resulting non-redundant repeat library was then classified with RepeatClassifier⁴⁴ (v2.0.1). To filter out protein-coding sequences, repetitive elements were mapped against a custom database, containing the annotated protein-coding sequences of Danio rerio, Seriola lalandi dorsalis, Cyclopterus lumpus, Xiphophorus couchianus, Oncorhynchus tshawytscha, and Salmo salar, using BLASTn⁵² (v2.9.0 + ) with an e-value threshold of 0.001. Any sequences matching protein-coding genes were removed from the library. The final repeat library was then used in RepeatMasker⁵³ (v4.1.1) to generate a genome report repeat content and to mask the genome for subsequent annotation with the BRAKER3 pipeline. RepeatMasker was run in standard mode, resulting in a hard-masked genome in which repetitive regions were replaced with ambiguous nucleotides prior to gene prediction.

A summary of repetitive sequence content in the Trachurus murphyi genome, including the relative contribution of major transposable element classes, their copy numbers, genomic distribution, and sequence divergence profiles, is provided in Table 4 and Fig. 3. A full subclassification of transposable elements with summation rows is provided in the Supplementary Table 1.

Table 4.

Repeat content and their proportion in the T. murphyi genome.

Type of repetitive element	Length (bp)	Percentage (%)
Retroelements	68,263,474	8.34
DNA transposons	72,717,968	8.89
Rolling-circles (Helitrons)	4,329,195	0.53
Unclassified repeats	100,161	0.01
Satellites	508,595	0.06
Simple repeats	45,640,029	5.58
Low complexity	3,683,752	0.45
Total repetitive sequences	195,243,174	23.86

Fig. 3.

Transposable elements (TE) in the genome of Trachurus murphyi. (A) Proportion of the assembled genome classified into major repeat categories based on RepeatMasker annotation. Percentages represent the faction of the total assembly length occupied by each repeat class. For percentage of each major TE categories, refers to Supplementary Table S1. (B) Hierarchical classification of transposable element families detected in the genome. Bars represent the number of TE copies per family, grouped into major TE categories (DNA transposons, LINEs, LTRs, PLE, rolling-circle transposons, and SINEs). Color intensity reflects copy number abundance. For the exact copy number per family, refers to Supplementary Table 1. (C) Repeat landscape showing the genomic proportion of major TE classes plotted against CpG-corrected Kimura divergence from their consensus sequences. The y-axis represents the percentage of the genome occupied by repeats within each divergence bin, and stacked colors indicate contributions from different TE classes. Lower divergence values correspond to more recent insertions, whereas higher values represent older elements. (D) Circos plot showing the chromosomal distribution of major TE classes across the 24 chromosomes. Repeat density was calculated in fixed genomic windows (500 kb), with radial axis values representing the proportion of each window occupied by the corresponding repeat class. Colors indicate DNA transposons (red), LINEs (yellow), LTRs (pink), SINEs (dark blue), and rolling-circle transposons (dark teal).

Analysis of transposable element divergence showed that 48.2% of annotated repeats exhibit low genetic distance (Kimura divergence ≤ 10%) from their respective consensus sequences (Fig. 3C). Among major element classes, LINEs, LTRs, and DNA transposons show enrichment of very low-divergence copies, with 18.9% of LINEs, 22.8% of LTRs, and 21.5% of DNA transposons displaying divergence values between 0 and 5%, consistent with relatively recent transposable element activity.

Gene prediction and functional annotation

To annotate protein-coding sequences in the Chilean jack mackerel genome, we used the BRAKER3⁵⁴ (v3.0.8) along with reference proteins from 18 species: Anabas testudineus (GCA_900324465.3), Carassius auratus (GCA_003368295.1), Clupea harengus (GCA_900700415.2), Danio rerio (GCA_000002035.4), Gadus morhua (GCA_902167405.1), Gasterosteus aculeatus (GCA_016920845.1), Oreochromis aureus (GCA_013358895.1), Oreochromis niloticus (GCA_001858045.3), Oryzias javanicus (GCA_003999625.1), Poecilia formosa (GCA_000485575.1), Salmo salar (GCA_905237065.2), Sander lucioperca (GCA_008315115.1), Scleropages formosus (GCA_900964775.1), Seriola dumerili (GCA_002260705.1), Seriola lalandi dorsalis (GCA_002814215.1), Sparus aurata (GCA_900880675.1), Trachurus trachurus (GCA_905171665.1), and Xiphophorus couchianus (GCA_001444195.1). All protein sequences were obtained from Ensembl Release 112. All reference proteomes were concatenated and used directly as input for BRAKER3 without prior manual curation, allowing the pipeline to internally select relevant protein evidence during gene prediction. It is worthy to mention that gene prediction was performed in protein-supported mode because RNA-seq transcriptomic data were not available, as high-quality total RNA suitable for sequencing could not be obtained due to technical limitations. Consequently, transcript-supported gene prediction was not performed.

Briefly, GeneMark-EPT⁵⁵ was used to identify genomic loci where extrinsic evidence support reliable gene predictions. These high-confidence predictions were then used to train the statistical model implemented in AUGUSTUS⁵⁶. Further refinements incorporated protein-derived extrinsic evidences from the reference proteomes listed above, with gene predictions subsequently optimized using TSEBRA⁵⁷. Functional annotation of the predicted protein-coding genes was conducted using DIAMOND⁵⁸ (v2.1.11) against the following public databases: NCBI non-redundant (nr), SwissProt, TrEMBL, and eggNOG. Protein domain searches were performed using Pfam with HMMER⁵⁹ v(3.3.2) and InterProScan⁶⁰ (v5.74.105). Gene Ontology (GO) terms were extracted from the InterProScan output, and KEGG Orthology (KO) annotation was conducted using KofamScan⁶¹ and the KOfam database.

Mitogenome assembly and annotation

To assemble the mitochondrial (mtDNA) genome of T. murphyi, we employed a strategy that utilizes the reference mitogenome of a closely related species from NCBI (accession number PP935111.1 in this case). This reference was used to capture mitochondrial PacBio HiFi reads, which were then assembled into a single circularized contig using MitoHiFi⁶² (v3.2.1). In addition to the annotation provided by MitoHiFi, the assembled mitogenome was further annotated using MITOS2 web server⁶³. To evaluate the depth of coverage, all PacBio HiFi reads were mapped back to the assembled mitogenome using minimap2⁶⁴ v(2.29). We used BLASTn⁵² (v2.10.1) to identify matches between the mitochondrial genome assembly and the nuclear genome assembly. Nuclear contigs showing > 99% sequence identity and shorter lengths than the mitochondrial genome assembly were filtered out from the mitochondrial genome assembly. The resulting mitogenome assembly and its annotations were visualized using the Proksee web tool⁶⁵. Mitochondrial protein-coding gene arrangements among Trachurus species were compared using MITOS2 GFF annotation data. Protein-coding genes were ordered according to their genomic coordinates and strand orientation, and gene arragements were compared across species to assess collinearity. Homologous genes were identified based on gene annotation names, and gene order conservation was evaluated by direct coordinate-based comparisons. As the analysis focused on structural gene arrangement rather than sequence divergence, no statistical tests were applied. The organization of the T. murphyi mitochondrial genome, read coverage across the assembled sequence, and collinearity with mitochondrial genomes from other Trachurus species are presented in Fig. 4.

Fig. 4.

Mitochondrial genome of Trachurus murphyi. (A) Circular representation of the T. murphyi mitochondrial genome (16,561 bp). Protein-coding genes (CDS), transfer RNAs (tRNAs), and ribosomal RNAs (rRNA) are indicated by distinct colors. Gene orientation is shown relative to the reference strand. Genomic coordinates are indicated in kilobases (kb). (B) Coverage depth across the mitochondrial genome based on mapped PacBio HiFi reads. The x-axis represents genomic position (bp), and the y-axis indicates read depth. Coverage is relatively uniform across the genome, with localized variation likely associated with repetitive or low-complexity elements. (C) Gene order comparison among mitochondrial genomes of six Trachurus species. Homologous genes are connected by black lines, illustrating collinearity of gene arrangement. The dendrogram indicates phylogenetic relationships among the species based on mitochondrial sequence similarity.

Data Records

The datasets generated in this study include raw sequencing reads, genome assemblies, and genome annotation files for Trachurus murphyi, have been deposited in public repositories such as NCBI and figshare⁶⁶. All sequencing data have been deposited in the NCBI Sequence Read Archive⁶⁷ database under BioProject accession PRJNA1420201 and BioSample accession SAMN55211191, including PacBio HiFi reads (SRR37159638-SRR37159640) and Omni-C reads (SRR37159637). The nuclear genome assembly has been deposited in GenBank⁶⁸ under accession GCA_055096975.1, and the mitochondrial genome under accession PZ051637.

Data Overview

The genomic datasets generated for Trachurus murphyi comprise a highly contiguous chromosome-level nuclear genome assembly and a complete mitochondrial genome. The nuclear genome assembly spans 818.321 Mbp across 532 scaffolds, with a scaffold N50 of 34.34 Mb, an N90 of 25.13 Mb, and a GC content of 42.31% (Table 1, Table 2). A total of 24 pseudo-chromosomes account for 95.80% of the assembled genome, with 34.31 Mb remaining as unplaced scaffolds and no ambiguous bases present (Table 2). BUSCO analysis indicated 99.0% completeness, including 97.9% single-copy, 1.1% duplicated genes, with 0.3% fragmented and 0.7% missing genes (Fig. 1D). Repetitive sequence comprises 195.24 Mb (23.86%) of the T. murphyi genome. Interspersed repeats account the majority of repetitive DNA, with retroelements representing 8.34% of the genome, DNA transposons 8.89%, rolling-circle transposons 0.53%, and unclassified repeats 0.01%. In addition, simple repeats comprise 5.58%, satellites 0.06%, and low-complexity regions 0.45% of the genome (Table 4). Gene annotation identified 47,734 predicted protein-coding genes with a total of 370,799 exons. The mean exon length is 189 bp (median = 126 bp), with 50% of exons ranging between 86 and 179 bp. These values are consistent with the typical exon size distribution observed in teleost genome⁶⁹. The average gene length is 7.0 kb, with a mean coding sequence (CDS) length of 1.35 kb. Genes contain an average of 7.8 exons and 6.8 introns (mean = 979 bp) (Fig. 5), consisting with gene structure characteristics observed in teleost genomes⁶⁹; functional annotation assigned homologous matches to 43,427 genes (90.98%) across public databases, with 9,080 genes supported by all eight annotation resources used (Fig. 5). The mitochondrial genome was assembled as a single circular contig of 16,561 bp, containing 13 protein-coding genes, 22 tRNA genes, and two rRNA genes, with PacBio HiFi reads mapping indicating uniform coverage across most of the genome (Fig. 4).

Fig. 5.

Characteristics and functional annotation of predicted protein-coding genes in Trachurus murphyi. (A) Violin plots showing the distribution of gene lengths and coding sequence (CDS) lengths (log10 scale). (B) Violin plot showing the distribution of exon counts per gene (log10 scale). (C) Distribution of exon and intron lengths (log10 scale). (D) UpSet plot summarizing functional annotation overlap among eight public databases (NCBI nr, SwissProt, TrEMBL, EGGnog, Pfam, InterProScan, GO, and KOfam). Vertical bars represent the number of genes in each intersection set, while connected dots below indicate the combination of databases contributing to each intersection. Horizontal bars on the right show the total number of genes annotated by each database.

Technical Validation

To evaluate the quality of our genome assembly and gene annotation, we employed several complementary approaches. First, we assessed potential contamination in the genome assembly using BlopToolKit³¹ (v4.4.3) to analyze GC content and sequencing depth. Contigs identified as contaminants were removed from the assembly (Table 1). After decontamination, we observed that nearly all GC content points clustered around 40%, indicating the absence of exogenous contamination from bacteria or other species (Fig. 1B). Second, we mapped the Omni-C reads to the assembled genome using BWA (https://github.com/lh3/bwa) to generate a Hi-C contact map. The resulting heatmap displayed high consistency across all chromosomes, providing strong evidence for the accurate sequencing, ordering, and orientation of contigs in the Chilean jack mackerel genome assembly (Fig. 1C). Third, we predicted centromeric and telomeric sequences in the T. murphyi genome and found that at least 12 chromosomes contained one of these characteristic chromosomal features (Fig. 2A, Table 2, Table 3).

Fourth, we evaluated assembly completeness using BUSCO⁷⁰ (v5.3.2). The analysis, based on the actinopterygii_odb10 database, revealed a completeness score of 99.0%, comprising 97.9% complete single-copy, 1.1% duplicated, 0.3% fragmented, and 0.7% missing genes (Fig. 1D). These results underscore the high completeness and accuracy of the Chilean jack mackerel genome assembly. Fifth, we assessed per-base consensus accuracy using Merqury⁷¹ (v1.3). The assembly shows very high consensus accuracy (QV = 52.8), corresponding to a base-level accuracy of 99.999%. The low number of missing k-mers (90 kb out of 818.321 Mb) indicates a nearly complete and high accurate assembly, with few sequence errors, indicative of a nearly error-free assembly. Sixth, we aligned PacBio HiFi reads to the assembled genome using minimap2⁶⁴ (v2.17), resulting in a mapping rate of 99.97%, further supporting the high consensus accuracy of the assembly. Seventh, we evaluated assembly contiguity using QUAST⁷² (v5.3.0), obtaining contig N50, scaffold N50, and additional key metrics (Table 5), all demonstrating the highly contiguous quality of the genome assembly.

Table 5.

QUAST evaluation of the T. murphyi genome assembly quality.

Assembly	Metric
# contigs ( >  = 0 bp)	523
# contigs ( >  = 1000 bp)	523
# contigs ( >  = 5000 bp)	523
# contigs ( >  = 10000 bp)	523
# contigs ( >  = 25000 bp)	523
# contigs ( >  = 50000 bp)	523
Total length ( >  = 0 bp)	818,310,266
Total length ( >  = 1000 bp)	818,310,266
Total length ( >  = 5000 bp)	818,310,266
Total length ( >  = 10000 bp)	818,303,047
Total length ( >  = 25000 bp)	814,740,251
Total length ( >  = 50000 bp)	807,049,109
# contigs	523
Largest contig	43,616,551
Total length	818,310,226
GC (%)	42.31
N50	34,335,234
N90	25,125,839
L50	11
L90	22

Eighth, we assessed the completeness of the predicted protein-coding sequences with BUSCO⁷⁰ (v5.3.2), achieving a BUSCO completeness score of 91.6% (S: 78.2%, D: 13.4%, F: 4.4%, M: 4.0%) for CDS and 91.7% (S: 78.3%, D: 13.4%, F: 4.7%, M: 3.6%) for proteins against the Actinopterygii database (Fig. 6). The predicted gene count (47,734 protein-coding genes) exceeds that reported for many diploid teleost genomes⁶⁹. To investigate whether this reflects unresolved haplotypic duplication or annotation artifacts, we assessed BUSCO duplication rates at the gene model level. BUSCO analysis of the predicted coding sequences and proteins recovered 13.4% duplicated BUSCOs, indicating redundancy within the annotated gene set. In contrast, BUSCO analysis of the genome assembly showed high completeness with low duplication (97.9% complete single-copy and 1.1% duplicated BUSCOs), suggesting that haplotypic duplication was largely resolved at the assembly stage. To assess the effect of protein evidence selection on gene prediction, an additional BRAKER3 run was performed using Trachurus trachurus (GCA_905171665.2) as a sole reference proteome, yielding 45,540 predicted protein-coding genes, comparable to the 47,734 genes obtained using multiple proteomes, indicating robustness of gene prediction to protein input choice. Detailed results of this analysis are provided in Supplementary Table 2. The elevated number of predicted protein-coding genes is likely attributable to annotation-related factors, including the absence of RNA-seq data, the use of multiple heterologous protein datasets for training, and the use of a hard-masked genome during gene prediction, which together may contribute to fragmented or redundant gene models. Therefore, the current annotation should be considered a preliminary gene set that may be refined in future studies incorporating transcriptomic data.

Fig. 6.

BUSCO completeness analysis of predicted protein-coding genes in the Trachurus murphyi. BUSCO assessment was performed using the Actinopterygii database (actinopterygii_odb10; n = 3,640 orthologs) Bars represent the percentage of Complete (including single-copy and duplicated), Fragmented, and Missing ortholgos for coding sequence (CDS) and translated protein predictions.

Finally, a total of 43,427 (90.98%) protein-coding genes were successfully annotated in at least one database, and 9,080 (19.02%) genes were supported by all eight databases used (Fig. 5D), indicating high-quality, accurate, and comprehensive annotation of the protein-coding genes in the T. murphyi genome. Taken together, all these lines of evidence confirm the accuracy, reliability, and high-quality of the Chilean jack mackerel genome assembly and annotation.

Author contributions

S.F.-F., C.C.-A., and F.A. conceived and designed the study. S.F.-F. and V.H.-Y. collected the Chilean jack mackerel samples and conducted laboratory procedures. C.Q. and F.A. conducted the bioinformatic analyses. F.A. drafted the initial version of the manuscript. All authors reviewed, revised, and approved the final manuscript.

Funding

This work was supported by funding provided by the Fondo de Investigación Pesquera y Acuicultura of Chile, FIPA N°2021-28 and FIPA N°2023-18 to SFF, Agencia Nacional de Investigación (ANID) grants: Fondecyt Regular 1220708 and Fondequip EQM200056 to FA. CC-A received partial support from ANID Fondecyt Regular grant 1251604 and ANID CIN250047.

Data availability

All raw sequencing reads, genome assemblies, and associated metadata are available through the National Center for Biotecnology Information (NCBI). Raw sequencing data are accessible via the Sequence Read Archive (SRA) under BioProject accession PRJNA1420201. The chromosome-level nuclear genome assembly and mitochondrial genome are available through GenBank under accessions GCA_055096975.1 and PZ051637, respectively. Genome annotation files generated in this study are available through Figshare at https://doi.org/10.6084/m9.figshare.31304044.

Code availability

No custom codes or script were used in this study. All bioinformatics tools and pipelines were executed following the instructions and manuals provided by the respective software developers. The specific software versions and corresponding parameters, where applicable, are detailed in the Methods subsection. Unless stated otherwise, default parameters were used for all analyses.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41597-026-07551-2.