Until recently, the family Callionymidae was placed into the order Perciformes, which is often considered a ‘polyphyletic taxonomic wastebasket for families not placed in other orders’ [1]. However, recent phylogenetic analyses suggest a placement of Callionymidae within the order Syngnathiformes, which currently contains ten families with highly derived morphological characters such as the pipefish and seahorses [1]. Syngnathiformes has recently been divided into two clades, a ‘long-snouted clade’ and a ‘benthic associated clade,’ each comprising five families [2]. The ‘long-snouted clade’ (Syngnathidae, Solenostomidae, Aulostomidae, Centriscidae, and Fistulariidae) is currently represented by genomes from the Gulf Pipefish (Syngnathus scovelli) and the Tiger Tail Seahorse (Hippocampus comes) [3, 4] and additional draft assemblies of pipefish [5]. A genome of the ‘benthic associated clade’ (Callionymidae, Draconettidae, Dactylopteridae, Mullidae, and Pegasidae) has not been sequenced and analysed yet. Callionymidae comprises 196 species [6], of which the common dragonet, Callionymus lyra (Linnaeus, 1758) (Figure 1), is one of three Callionymus species inhabiting the North Sea [7]. All three species also occur in the East Atlantic, and the Mediterranean Sea [6]. They represent essential prey fish for commercially important fish species such as the cod (Gadus morhua) [8]. The males of the North Sea dragonet species (C. lyra, C. maculatus, C. reticulatus) show strong morphological differentiation in the form of species-specific colouration and size relations. The much less conspicuous females can be distinguished morphologically, with rather high inaccuracy, by the presence or absence of their preopercular, basal spine and by various percentual length ratios. The great resemblance among the different species’ females, together with the fact that all three species can be found in sympatry, suggests there is the possibility of hybridization among them.

Here, we present the chromosome-level genome of the common dragonet, representing the first genome of the ‘benthic associated’ Syngnathiformes clade as a reference for future population genomic, phylogenomic, and comparative genomic analyses. The chromosome-level genome assembly was generated as part of a six-week university master’s course. For a detailed description and outline of the course, see Prost et al. [9].

We sampled two Callionymus lyra (NCBI: txid34785; Fishbase ID:23) individuals (one of each sex) during a yearly monitoring expedition to the Dogger Bank in the North Sea (Female: 54° 59.189′ N 1° 37.586′ E; Male: 54° 48.271′ N 1° 25.077′ E) with the permission of the Maritime Policy Unit of the UK Foreign and Commonwealth Office in 2019. The samples were initially frozen at −20 °C on the ship and later stored at −80 °C until further processing. The study was conducted in compliance with the ‘Nagoya Protocol on Access to Genetic Resources and the Fair and Equitable Sharing of Benefits Arising from Their Utilization’.

We extracted high molecular weight genomic DNA (hmwDNA) from muscle tissue of the female individual following the protocol by Mayjonade et al. [10]. Quantity and quality of the DNA was evaluated using the Genomic DNA ScreenTape on the Agilent 2200 TapeStation system (Agilent Technologies). Library preparation for long-read sequencing followed the associated protocols for Oxford Nanopore Technologies (ONT, Oxford, UK) Rapid Sequencing Kit (SQK-RAD004). A total of seven sequencing runs were performed using individual flow cells (FLO-MIN106 v.9.41) on a ONT MinION v.Mk1B.

Additionally, we sent tissue samples to BGI Genomics (Shenzhen, China) to generate additional sequencing data. A 100 bp paired-end short-read genomic DNA sequencing library was prepared from the muscle tissue of the female individual. This library was later used for genome assembly polishing. Moreover, a 100 bp paired-end RNAseq library was prepared for pooled RNA isolates derived from kidney, liver, gill, gonad, and brain tissues of the male individual. Both libraries were sequenced on BGI’s DNBseq platform (BGISEQ-500/DNBSEQ-G50 sequencing) [11]. We received a total of 159,925,221 read pairs (∼32 Gbp) of pre-filtered genomic DNA sequencing data and 61,496,990 read-pairs (∼12.3 Gbp) of pre-filtered RNAseq data.

Furthermore, we prepared a Hi-C library using the Dovetail™ Hi-C Kit (Dovetail Genomics, Santa Cruz, California, USA) from muscle tissue of the female and sent the library to Novogene Co., Ltd. (Beijing, China) for sequencing on an Illumina NovaSeq 6000. Sequencing yielded a total of 104,668,356 pre-filtered 150 bp paired-end read pairs or 31.4 Gbp of sequencing data. This data was used for proximity-ligation scaffolding of the assembly.

We estimated the genome size for C. lyra using both k-mer frequencies and flow cytometry. The k-mer frequency for K = 21 was calculated from the short-read DNBseq data and summarized as histograms with jellyfish v.2.2.10 (RRID:SCR_005491) [12]. Plotting the histograms and calculating the genome size and heterozygosity with GenomeScope v.1.0 (RRID:SCR_017014) [13] resulted in a genome size estimate of approximately 562 Mbp. For the genome size estimation using flow cytometry, frozen muscle tissue was finely chopped with a razor blade in 200 μl LeukoSure Lyse Reagent (Beckman Coulter Inc., Fullerton, CA, USA). Large debris was removed by filtering through a 40 μm Nylon cell strainer and an RNAse treatment was performed with a final concentration of 0.3 mg/ml. Simultaneously, we stained the DNA in the nuclei with propidium iodide (PI) at a final concentration of 0.025 mg/ml and incubated the solution for 30 min at room temperature, protected from light exposure. Fluorescence intensities of the nuclei were recorded on the CytoFLEX Flow Cytometer (Beckman Coulter Inc., Fullerton, CA, USA). The domestic cricket (Acheta domesticus, C-value: 2.0 pg) was used as a reference to determine the genome size of C. lyra. For a more precise estimate we analysed five independent technical replicates resulting in an average C-value of 0.66 pg, which corresponds to a haploid genome size of approximately 645 Mbp.

Nanopore raw signal data (fast5) of the seven sequencing runs were base-called with Guppy v.3.2.4 (ONT) using the high accuracy setting. All individual sequencing runs were examined and compared with NanoComp v.1.0.0 [15] (Figure in GigaDB [14], Table 1).

The final dataset, after concatenation of all read-files, was further examined with NanoPlot v.1.0.0 (Table 1) [15]. Concatenation of all read-files resulted in a total dataset of 31 Gbp or approximately 55-fold coverage as the basis for the genome assembly.

We assembled the genome of C. lyra with wtdbg2 v.2.2 (RRID:SCR_017225) [16] using the default parameters for ONT reads. The resulting assembly was subjected to a three-step polishing approach. First, a single iteration of racon v.1.4.3 (RRID:SCR_017642) [17] corrected for errors typical of the MinION platform: homopolymers and repeat errors. Next, we used one iteration of medaka v.0.11.5 [18] on the racon-polished assembly. According to the developers medaka is most effective after a polishing run with racon. Following polishing with the long-read data, we used three iterations of pilon v.1.23 (RRID:SCR_014731) [19] to correct for random errors and single-base errors with the high-quality short-read data.

We calculated assembly continuity statistics using QUAST v.5.0.2 (RRID:SCR_001228) [20] and performed a gene set completeness analysis using BUSCO v.4.0.6 (RRID:SCR_015008) [21] with the provided database for Actinopterygii orthologous genes (actinopterygii_odb10). The final polished assembly had 1,782 contigs and a total length of 569 Mbp, which is marginally larger than the k-mer based estimate of 562 Mbp and 84 Mbp shorter than the flow cytometry estimate. This is expected, because very repetitive regions are usually missing or collapsed in a genome assembly, which could explain the shorter assembly length compared to the flow cytometry size estimate. The assembly shows a high continuity with long contigs of up to 10.7 Mbp and a contig N50 of >2.2 Mbp (Table 1). The genome assembly completeness analysis identified 95.0% complete BUSCO genes (93.6% complete, single copy) and only 4.4% missing BUSCOs, which suggests that the assembly contains most of the coding regions of the genome (Figure 2, Table 2).

To achieve chromosome-length scaffolds, we used the long-read based assembly and the generated Hi-C data as input for the HiRise scaffolding pipeline [22] as part of the Dovetail Genomics’ scaffolding service. HiRise made 538 joins and 10 breaks resulting in a scaffolded assembly with a total of 1,254 scaffolds and a scaffold N50 of 26.7 Mbp. Over 94.5% (538 Mbp) of the total assembly length was scaffolded into 19 chromosome-length scaffolds (Figure 3A). The number of chromosome-length scaffolds is consistent with the haploid number of chromosomes derived from karyotypes of females of two Callionymidae species (C. beniteguri and Repomucenus ornatipinnis) [23]. Therefore, the number of chromosomes appears to be relatively conserved within Callionymidae and it is likely that C. lyra follows the same chromosomal sex determination system as C. beniteguri and R. ornatipinnis (♀: X1X2–X1X2 (2n = 38); ♂: X1X2–Y (2n = 37)) [23].

For a final assembly quality control, we mapped the raw nanopore reads with minimap2 v.2.17-r941 [24] and the DNBSeq data with bwa-mem v.0.7.17-r1194-dirty (RRID:SCR_010910) [25] onto the final assembly with a high mapping rate of 94.8% and 98.62%, respectively. We further checked the assembly for contamination with BlobTools v.1.1.1 (RRID:SCR_017618) [26]. This analysis identified minor contamination from Proteobacteria (26 short contigs, in total 0.25 Mbp) and Uroviricota (2 contigs, in total 0.12 Mbp) (Figure 4). No contamination was found in the 19 chromosome-length scaffolds. Subsequently, we removed all contaminations and contigs with a length of <200 bp from the final assembly (for final statistics see Table 3). In addition, we screened for mitochondrial sequence contamination with BLASTN v.2.9.0+ (RRID:SCR_001598) [27] using the available mitochondrial genome sequence of C. lyra (Accession No.: MN122938.1) as a reference. A single sequence of mitochondrial origin (169 bp) was identified on one scaffold. This partial mitochondrial sequence could either be an assembly artifact or nuclear mitochondrial DNA (numt).

A synteny plot comparing the polished wtdbg2 contig assembly with the final chromosome-level assembly, generated with JupiterPlot v.1.0 [28], found overall strong agreements with only few differences (Figure 3B). These likely constitute assembly errors in the contig assembly that were fixed by HiRise during scaffolding. A BUSCO analysis of the final assembly found slightly less complete BUSCO genes compared to the wdtbg2 contig assembly (94.5% vs. 95.0%) (Figure 2, Table 2).

In addition to the genome, we assembled the transcriptome of C. lyra for subsequent use in the genome annotation using Trinity v.2.9.0 (RRID:SCR_013048) [29, 30] based on the 12.3 Gbp multi-tissue RNAseq data. The resulting transcriptome assembly has a total length of 255.5 Mbp (Table 3). BUSCO analysis suggests a high transcriptome completeness with 87.8% of orthologous genes found in the transcriptome assembly (Figure 2, Table 2).

Here we report the first genome assembly of the ‘benthic associated’ Syngnathiformes clade, the sister group to the ‘long-snouted clade’ (e.g., seahorses and pipefish). The annotated genome of Callionymus lyra, with its high continuity (chromosome-level), provides an essential reference to study speciation and potential hybridization in Callionymidae and is an important resource for phylogenomic analyses among syngnathiform fish.