Background

Sex determination in fish is diverse and shows high plasticity, with different underlying mechanisms among species. Sex determination mechanisms in fish include genetic sex determination and environmental sex determination as well as joint regulation by genetic and environmental factors [1]. Fishes have almost all types of sex determination systems found in vertebrates. The two main types of sex determination systems in fish are XX/XY and ZZ/ZW. Most fish have homozygous sex chromosomes that cannot be distinguished by shape [2]. With recent advances in high-throughput sequencing technologies and bioinformatics methods, it has become possible to identify and characterize sex chromosomes with low divergence. Male and female chromosomes in genome assemblies have been obtained in a variety of fish.

The family Clariidae (airbreathing catfishes) in the class Actinopteri and order Siluriformes is widely distributed in freshwater regions of Africa and Asia. Clariidae consists of 117 species in 16 genera (https://www.fishbase.in/search.php). The genus Clarias comprising 61 species is the largest genus in Clariidae. Clarias is an excellent model for understanding the evolution of sex determination because the genus includes species with an unusual diversity of sex chromosome systems. For example, Clarias anguillaris and Clarias ebriensis have the ZZ/ZW sex determination system [3], Clarias macrocephalus and Clarias fuscus have the XX/XY sex determination system [4, 5], and Clarias gariepinus have XX/XY or ZZ/ZW sex determination system in different populations [6]. Clarias batrachus, C. fuscus, C. macrocephalus, and C. gariepinus are economically important species in Asian aquaculture (Food and Agriculture Organization of the United Nations, https://doi.org/10.4060/ca9229en). Furthermore, C. fuscus and C. gariepinus are sexually dimorphic, with differences in growth rates between males and females [7, 8]. Therefore, it is necessary to assemble chromosome-level genomes of these species to provide a basis for understanding the evolution of sex chromosomes of catfishes.

Hong Kong catfish (Clarias fuscus), characterized by high adaptability, high nutritional value, and tender flesh, is an economically valuable freshwater fish species and is widely cultivated in southern China [9, 10]. It shows asynchronous gonadal development as well as multiple spawning, with a high reproductive capacity, age of sexual maturity of 10–12 months, and the ability to reproduce 3–4 times a year [11, 12]. Hong Kong catfish exhibit sexual dimorphism, particularly in body size. Under the same breeding conditions, the growth rate of Hong Kong catfish is significantly higher in males than in females [8]. When farmed Hong Kong catfish reach market size, females are in the gonadal development stage. Female abdomens are enlarged, and their gonads account for a much higher proportion of body weight than males, resulting in lower prices for females than for equally sized males. The monosex fish culture of Hong Kong catfish has important economic significance. The lack of genomic resources for this species is a serious impediment to monosex breeding. Thus, it is necessary to obtain the chromosome-level genome of the Hong Kong catfish as a resource for molecular breeding.

The genomes of Clariidae have been published for Clarias batrachus [13], Clarias macrocephalus [14] and Clarias magur [15]. The genomes of Clarias fuscus, a common catfish in Southeast Asia, have not been reported. Moreover, large-scale genomic analyses at the chromosome level were not well characterized in Clarias due to the fragmented assembly of C. batrachus, C. macrocephalus and C. magur genomes. In this study, the first high-quality chromosome-level genome of Hong Kong catfish was constructed by PacBio sequencing and high-throughput chromosome conformation capture (Hi-C) technology, and sex-determining candidate regions and sex- related genes were identified by sex-linked QTL.

Results

Genome sequencing and assembly

A total of 179.58 Gb of subreads base was obtained from the PacBio sequencing library, including 10,269,296 subreads with an average length of 17,488 bp and subread N50 of 25,523 bp. A total of 107.18 Gb of clean data was obtained from the HI-C sequencing library, including 725,685,598 clean reads with a Q20 of 96.7% and Q30 of 88.3%. A total of 56.85 Gb of clean data was obtained from the small-fragment genomic library, including 379,011,716 clean reads, with a Q20 of 95.8% and Q30 of 87% (Supplementary Table S1).

PacBio data were used to construct the preliminary genome assembly. The size of the preliminary genome assembly was 966.34 Mb, including 897 contigs, and the contig N50 was 8.18 Mb. Interrupting erroneous contigs in the preliminary genome based on HI-C data (957 contigs formed in total). The interrupted contigs were sorted to obtain chromosomal level genes. Finally, the Hong Kong catfish chromosome-level genome assembly was obtained with a length of 933.40 Mb, 28 chromosomes (containing 416 contigs), contig N50 = 8.52 Mb, and scaffold N50 = 35.68 Mb (Table 1, Supplementary Table S2). According to the BUSCO results, the genome contained 2,493 (96.40%) complete BUSCOs, including 2,405 single-copy BUSCOs and 88 duplicate BUSCOs (Supplementary Table S3). The results indicate that the genome assembly has high coverage and completeness.

Table 1 Genome assembly statistics for the Clarias fuscus genome
Full size table

Genome annotation

A total of 23,345 protein-coding genes were predicted in the Hong Kong catfish genome (Table 2; Fig. 1), with an average length of 19,425.94 bp. Of these, 18,493 genes encoded single transcripts and 4,852 genes encoded multiple transcripts, for a total of 32,216 transcripts. The characteristics of functional genes (length distribution of gene, coding sequence, exon and intron) were compared with those of other catadromous fishes, and the gene characteristics of Hong Kong catfish was similar to that of other Siluriformes (Supplementary Figure S1). The genes were annotated using the NR, TrEMBL, SwissProt, InterPro, KEGG, and GO databases. A total of 22,009 genes were annotated, representing 94.28% of all genes (Supplementary Table S4).

Fig. 1
figure 1

Genome characteristics of C. fuscus. From the outer to inner circle: gene distribution, repeat sequence distribution, tandem repeat sequence distribution, DNA transposon distribution, long terminal repeated (LTE) distribution, long interspersed nuclear element (LINE) distribution, short interspersed nuclear element (SINE) distribution and non-coding RNA (ncRNA) distribution. Bar height is proportional to number of items mapped to each genomic position

Full size image
Table 2 Statistical summary of protein-coding gene prediction in the Clarias fuscus genome
Full size table

The duplicated sequences accounted for 53.50% of the Hong Kong catfish genome (Fig. 1). DNA transposons (30.23%) were the most common TE type in the Hong Kong catfish genomic, followed by long terminal repeats (LTR, 20.27%) and long interspersed nuclear elements (LINEs, 10.14%) (Supplementary Table S5, Supplementary Table S6).

tRNAscan-SE, BLASTN, and Rfam were used to predict noncoding RNA, and a total of 12,110 tRNAs, 1,672 rRNAs, 506 snRNAs, and 278 miRNAs were predicted in the Hong Kong catfish genome (Fig. 1, Supplementary Table S7).

Genomic evolution analysis

Gene family clustering was performed based on protein sequences of Hong Kong catfish. A total of 15,977 gene families and 3,185 single-copy genes were identified (Supplementary Figure S2, Supplementary Table S8). The gene families of 16 fishes (Clarias fuscus, Latimeria chalumnae, Lepisosteus oculatus, Danio rerio, Electrophorus electricus, Astyanax mexicanus, C. magur, Silurus meridionalis, Pangasianodon hypophthalmus, Ictalurus punctatus, Pelteobagrus fulvidraco, Esox lucius, Gadus morhua, Oryzias latipes, Oreochromis niloticus, and Takifugu rubripes) were compared. A total of 60 unique, 353 expanded and 851 contracted gene families were identified in Hong Kong catfish (Supplementary Table S9).

Phylogenetic analyses showed that Hong Kong catfish and C. magur are clustered in a single branch. In addition, the Hong Kong catfish is closely related to S. meridionalis, P. hypophthalmus, I. punctatus, and P. fulvidraco, which belong to the order Siluriformes. Furthermore, Hong Kong catfish diverged from its most closely related species, C. magur, about 63.7 (60.7–66.5) million years ago (Fig. 2).

Fig. 2
figure 2

Phylogenetic analysis of 16 fishes. At each node, the predicted divergence time (million years ago) is marked. The green number on each branch represents the number of expanded gene families, and red indicates the number of contracted gene families

Full size image

All 28 chromosomes of Hong Kong catfish were smaller than the 29 chromosomes of I. punctatus and 30 chromosomes of P. hypophthalmus (Fig. 3). In comparison with the two species of catfish, Hong Kong catfish Chr1, 9, 12, and 14 underwent chromosomal breakage and recombination events during evolution. In addition, chromosome breakage and recombination events were found on Chr3 and Chr5 in comparison with the I. punctatus chromosomes.

Fig. 3
figure 3

Collinearity analysis of C. fuscus and other teleost genome. (A) Collinearity analysis of C. fuscus and I. punctatus genomes. Blue and orange lines represent the chromosomes of C. fuscus and I. punctatus, respectively. (B) Collinearity analysis of C. fuscus and P. hypophthalmus genomes. Blue and orange lines represent the chromosomes of C. fuscus and P. hypophthalmus, respectively

Full size image

Identification of the sex-determination region

Each LG of the genetic linkage map corresponded to one chromosome of the genome assembly (Table 3). According to the mapping results, the sex-linked QTL was located on Chr13 of the genome assembly. The QTL (33.64–176.02 cM) corresponded to a 30.26 Mb region (from 0.02 to 30.28 Mb) on the genome (Fig. 4). Based on the gene annotation results for chr13, 785 genes were identified in the sex-linked QTL (Supplementary Table S10). Among the 785 genes in the sex-linked QTL, 18 sex related genes were identified, including fanconi anemia group m protein (fancm), EF-hand calcium-binding domain-containing protein 2 (efcab2), SRA stem-loop-interacting RNA-binding protein (slirp), F-box protein 30 (fbxo30), F-box protein 34 (fbxo34), bone morphogenetic protein 2 (bmp2), and AT-rich interactive domain-containing protein 4a (arid4a), transforming growth factor-beta receptor-associated protein 1 (tgfbrap1), adenylate kinase 7 (ak7), protein phosphatase 2 regulatory subunit B’’gamma (ppp2r3c), fibronectin type III and ankyrin repeat domains 1 (fank1), akt serine/threonine kinase 1 (akt1), muts homolog 4 (msh4), estrogen-related receptor b (esrrb), progesterone receptor membrane component 2 (pgrmc2), cytochrome P450 26B1 (cyp26b1), gremlin 2 (grem2), and spermatogenesis associated 17 (spata17) (Table 4).

Fig. 4
figure 4

Location of sex-related genes in the QTL interval of Clarias fuscus. The lower abscissa is the genetic distance, in centimorgans (cM); the upper abscissa is the genomic position, in megabases (Mb); the ordinate is the LOD threshold; black dots are SNP markers; brown dots are sex-related genes

Full size image
Table 3 Correspondence between the physical genome and genetic linkage map of Clarias fuscus
Full size table
Table 4 Sex-related genes in the sex-linked QTL region of Clarias fuscus
Full size table

Discussion

Some fish species exhibit sexual growth dimorphism, including significant differences in body size and growth rate between males and females. Thus, monosex culture of the sex with dominant growth will greatly improve the economic value of fish species. For example, in Cyprinus carpio, Oncorhynchus mykiss, and Cynoglossus semilaevis, female fish grow faster than males [16,17,18]. In contrast, some fish males grow faster than females, such as C. fuscus, P. fulvidraco, Oreochromis aureus, and I. punctatus [19,20,21]. Therefore, the assembly of a high-level reference genome and screening of sex-related genes are of great significance for the development of the single-sex aquaculture industry for catfish and clarifies the sex determination mechanism in the species.

In the present study, the high-quality chromosome-level genome of Hong Kong catfish (C. fuscus) was assembled. The Hong Kong catfish chromosome-level genome assembly length was 933.40 Mb and scaffold N50 = 35.68 Mb. In a comparative analysis of Siluriformes [13,14,15, 22,23,24,25], the genome size was smaller than that of C. magur (941 Mb) and larger than those of C. macrocephalus (883 Mb), C. batrachus (821.8 Mb), I. punctatus (783 Mb), and P. fulvidraco (732.8 Mb). The scaffold N50 values (35.68 Mb) were higher than those of C. magur (1.3 Mb), C. macrocephalus (80.8 kb), C. batrachus (361.2 kb), I. punctatus (7.73 Mb), and P. fulvidraco (25.8 Mb). These results show that the construction of a high-quality Hong Kong catfish genome assembly in this study. The Hong Kong catfish genome assembly was anchored to 28 chromosomes, which is consistent with previously reported karyotype data [26]. A total of 23,345 functional genes and 53.50% repetitive sequences were identified by annotation of the genome of Hong Kong catfish. In addition, comparing the repeated sequences of Hong Kong catfish with other Siluriformes, which was higher than estimates for C. magur (43.72%) [15], P. fulvidraco (43.31%) [27], I. punctatus (41.1%) [28], C. macrocephalus (38.28%) [14] and C. batrachus (30.3%) [13].

Previous studies of sex-linked QTL mapping have identified a sex-linked QTL in Hong Kong catfish in LG13 (Lin et al., 2022). By comparing the SNP markers of the genetic map with the Hong Kong catfish genome, LG13 was found to correspond to Chr13 in this study. This QTL interval occupied a 30.26 Mb region of Chr13 (ranging from 0.02 to 30.28 Mb) and contained 785 genes. The total length of Chr13 was 33.45 Mb, and the QTL region occupied 90.33% of the chromosome. It is speculated that Chr13 is the sex chromosome of Hong Kong catfish. Sex chromosomes have been identified in some Siluriformes, such as I. punctatus, P. fulvidraco, S. meridionalis and P. hypophthalmus [29,30,31,32]. Based on the identification of sex chromosomes, candidate sex-determining genes have been identified, such as breast cancer anti-estrogen resistance protein 1 (bcar1) in I. punctatus, PDZ domain-containing gene (pfpdz1) in P. fulvidraco, anti-Müllerian hormone receptor type 2 on the Y chromosome (amhr2y) in S. meridionalis and male-specific duplication of amhr2 (amhr2by) in P. hypophthalmus [29,30,31,32]. Based on the genome comparison results of the Hong Kong catfish and other fishes (Fig. 3), the bcar1 of I. punctatus is located on chromosome 4 [29], which compared to chromosome 7 of the Hong Kong catfish. The amhr2by of P. hypophthalmus is located on chromosome 4 [32], which compared to chromosome 2 of Hong Kong catfish. The sex-determining regions and genes of the above two Siluriformes fishes were not mapped to Chr13 of the Hong Kong catfish. This may be due to the fact that the genomes of Siluriformes fishes are generally large and multiple genome duplication events have occurred. During the replication process, the sex-determination-related mechanism of Siluriformes fishes also changed, thus presenting a variety of inconsistent sex-determination genes in Siluriformes fishes. In summary, the identification of sex chromosomes in Hong Kong catfish provides a basis for exploring the sex determination mechanism and sex determination genes.

Based on the gene annotation results for chr13, 18 sex related genes were identified among 785 genes in the sex-linked QTL. Among the 18 sex related genes, five infertility-related genes were identified, including fancm, slirp, arid4a, ak7, and akt1. Hypogonadism have been noted in both male and female mice with a fancm deficiency [33], and a homozygous fancm frameshift pathogenic variant causes male infertility in humans [34]. The slirp gene regulates male fertility, and its loss of function will alter the sperm structure and mitochondrial morphology [35]. arid4a regulates male fertility. Mice lacking arid4a and arid4b (Arid4a(−/−) and Arid4b(+/−)) exhibit the progressive loss of male fertility with hypogonadism and spermatophore hypoplasia [36]. The homozygous missense mutation L673P in ak7 leads to primary male infertility and multiple morphological anomalies of the flagella [37]. Female akt1(-/-) mice display reduced fertility and abnormal estrous cyclicity [72].

In addition, the remaining 13 sex-related genes were associated with gametogenesis, including efcab2, ppp2r3c, fank1, cyp26b1, spata17, bmp2, tgfbrap1, pgrmc2, grem2, fbxo30, fbxo34 and esrrb. The efcab2 gene was detected only in mouse tissues of the testis and may be involved in the control of sperm flagellar motility. In addition, it was specifically localized in spermatogenic cells from primary spermatocytes to elongate spermatids within the seminiferous epithelium; however, neither spermatogonia nor somatic cells were expressed [38]. The PPP2R3C protein is involved in the ontogeny of multiple organs and is especially critical for testis development and spermatogenesis [68]. PPP2R3C acts as a regulator of the PP2A and PP5 phosphatase holoenzymes and may be critical in the early signaling cascade controlling human sex determination [69]. fank1 is specifically expressed in the testis from the meiotic to haploid stages of spermatogenesis and may play a key role in spermatogenesis by functioning as a transcription factor [39]. Additionally, fank1-knockout mice have reduced sperm counts and increased apoptotic spermatocytes, which are mainly spermatogonia and spermatocytes [40]. cyp26b1 maintains low levels of retinoic acid in the developing testis, thereby preventing entry into meiosis and acting as a survival factor to prevent apoptosis in male germ cells [41]. The overexpression of SPATA17 leads to accelerated apoptosis in a zebrafish spermatogonial cell line [42]. bmp2 is a member of the TGF-β superfamily and is expressed in an ovary-specific manner during early gonadal development [43]. Overexpression of the tgfbrap1 gene suppresses granulosa cell E2 and P4 secretion, while tgfbrap1 knockdown enhances E2 and P4 secretion, suggesting that tgfbrap1 regulates apoptosis in goose follicle granulosa cells [44]. pgrmc2 may regulate the maturation of zebrafish oocytes by regulating receptors and steroids in the ovary [45]. Female grem2-knockout mice have irregular fertility and estrous cycles accompanied by a significant reduction in ovarian anti-Mullerian hormone production by growing follicles [46]. The fbxo30 gene regulates chromosome segregation during oocyte meiosis [47]. fbxo34 regulates the G2/M transition and late entry of meiotic oocytes [48]. msh4 is an important gene involved in meiosis, and mutations in this gene may be associated with female infertility and male non-obstructive azoospermia [49]. esrrb functions upstream of bmp4 (bone morphogenetic protein 4) in the extraembryonic ectoderm, regulating primordial germ cell development in mice [50]. The discovery of the above sex-related genes provides methodological guidance and an important tool for breeding techniques aimed at Hong Kong catfish sex control and lays the foundation for subsequent studies of mechanisms underlying sex determination.

In this study, the chromosome-level genome and sex-linked loci of Hong Kong catfish provide a research basis for subsequent developmental sex-linkage markers.Subsequently, the genetic male juvenile Hong Kong catfish were identified by sex-linked markers, and the biological sex of the genetic male Hong Kong catfish was changed by artificial sex reversal. So as to obtain the Hong Kong catfish of genetic male and physiological female. Then, sex-reversed Hong Kong catfish and normal male Hong Kong catfish were bred as parents, and YY Hong Kong catfish was sex-linked markers by sex-linked markers. Finally, based on YY Hong Kong catfish, the family groups of single-males Hong Kong catfish can be produced. In addition, sex-linked loci also provide candidate regions for identifying sex-determining genes. Subsequently, the sex-determining gene of Hong Kong catfish was knocked out by gene knockout technology, and a single-sex Hong Kong catfish population could be produced. In summary, the chromosome-level genome and sex-linked loci of the Hong Kong catfish can promote the development of the monosex culture industry of the Hong Kong catfish.

Conclusions

The chromosomal-level genome assembly of Hong Kong catfish was determined. The continuity and completeness of the Hong Kong catfish genome were consistent with those of other high-quality teleost fish genomes. Accordingly, these data provide a useful reference for systems biology and comparative genome evolution analyses. A sex-linked QTL was focused on Chr13, which was predicted as the Y chromosome, and eighteen genes in this region were considered as sex-related genes. The newly established reference genome provides an important bases for aquaculture and artificial breeding of Hong Kong catfish.

Methods

Sample preparation

Hong Kong catfish were obtained from the Guangxi Introduction and Breeding Center of Aquaculture (Nanning, China). One male (body weight 180.53 g, body length 26.8 cm) and one female (body weight 167.18 g, body length 25.5 cm) Hong Kong catfish of similar sizes were collected. After anesthesia in a bath of eugenol (1:10,000), dissection was performed after decapitation, and the mid-section of gonadal tissue was cut from each individual and stored at 4 °C in Bouin’s Fixative Solution. The head kidney, muscle, brain, liver, gonads (testis and ovaries), kidney, heart, stomach, spleen, and skin tissue were collected, snap-frozen in liquid nitrogen, and stored at -80 °C. Genomic DNA was extracted from muscle tissue of male fish using the CTAB method. The concentration and quality were examined using a NanoDrop 2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and 0.8% agarose gel electrophoresis. Total RNA was extracted from each tissue type of male and female Hong Kong catfish using the TRIzol method. The RNA purity and integrity were examined using a NanoDrop 2000 spectrophotometer (NanoDrop Technologies) and a Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA, USA). The extent of RNA degradation was examined by 1.5% agarose gel electrophoresis.

Library construction and sequencing

Genomic PacBio sequencing libraries were constructed using the SMRTbell Express Template Prep Kit 2.0 (Pacific Biosciences, Menlo Park, CA, USA) according to the manufacturer’s instructions. Sequencing libraries were used to determine the concentration and fragment size distribution of samples using the FEMTO Pulse System (Agilent) and Qubit 3.0 fluorometer (Life Technologies, Carlsbad, CA, USA). Fragment size selection was performed using the BluePippin system (Sage Science, Beverly, MA, USA) to remove sequenced fragments below 25 kb. After the libraries were tested, sequencing was performed on the PacBio Sequel II platform (Pacific Biosciences). For the construction of the Hi-C library, 1 g of muscle tissue was used to prepare a library according to previously established protocols [51]. The Hi-C library was sequenced on the MGI-SEQ 2000 platform (BGI, China). Clean data were obtained using HTQC (v1.92.310) [52] for quality control.

Small-fragment genomic libraries were constructed using the VAHTS Universal DNA Library Prep Kit for MGI (Vazyme, Nanjing, China) according to the manufacturer’s recommendations. After library construction, the concentration and fragment size distribution of samples were determined using a Qubit 3.0 fluorometer (Life Technologies) and Bioanalyzer 2100 system (Agilent Technologies). After the libraries were tested, sequencing was performed on the MGI-SEQ 2000 platform. HTQC (v1.92.310) [52] was used to remove adapters, duplicate reads, N-containing sequences (≥ 10%), and low-quality (Qphred ≤ 5) reads from the raw data.

Whole tissue mRNAs from female and male fish were used to construct transcriptome libraries for male and female fish, respectively. The libraries were constructed using the VAHTS Universal V6 RNA-seq Library Kit for MGI (Vazyme) according to the manufacturer’s instructions. Library quality and size were determined using a Qubit 3.0 Fluorometer (Life Technologies) and a Bioanalyzer 2100 system (Agilent Technologies). Sequencing was performed on the MGI-SEQ 2000 platform. SOAPnuke (v2.1.0) [53] was used to remove adapters, N-containing sequences (≥ 0.5%), and low-quality (Qphred ≤ 20) reads.

Genome assembly

A preliminary genome assembly was obtained using Mecat (v2.0) [54] for PacBio sequencing. After the preliminary assembly was completed, error correction was performed using SMRT Link (v8.0). Then, the small-fragment data were used for genome polishing using pilon (v1.22) [55]. Finally, the contig-level genome assembly was obtained. The genomic HI-C sequencing data were compared using BWA-mem (v.0.7.16a-r1181) [56], and single-end reads and sequences beyond 500 bp from the enzyme cut site were removed. Filtered Hi-C data were used for chromosome construction. The contigs were clustered, sorted, and oriented using 3D-DNA [57] to obtain chromosome-level genomes. The chromosome-level genome was visualized and error-corrected using JuiceBox [58]. The completeness of the Hong Kong catfish genome assembly was assessed using BUSCO (v3.0.2) [59] based on the single-copy homologous gene set (vertebrata_odb9) in OrthoDB.

Genome prediction and annotation

Homology-based annotation and de novo annotation methods were used to identify repetitive sequences in the genome. First, transposable element (TE) sequences were searched from the Repbase (v.21.01) [60] database based on homology using RepeatMasker (v.4.09) and RepeatProteinMask (v.4.09) [61]. Next, RepeatModeler (v.1.0.11) [62] and LTRfinder (v1.0.5) [63] were used to construct the Hong Kong catfish repeat sequence database de novo. Repeat sequences were then identified from the constructed database using RepeatMasker (v.4.09). In addition, TRF [64] was used to identify tandem repeat sequences. Finally, the results obtained by homology-based annotation and de novo annotation were integrated, and the final annotation results of duplicate sequences were obtained by removing the non-redundant parts after overlapping.

The homology annotation, de novo annotation, and transcriptome annotation were used to predict the structure and function of protein-coding genes. The coding-gene annotation information of C. magur, P. hypophthalmus, S. meridionalis, P. fulvidraco, and I. punctatus were selected for homology-based annotation. Genomes of these species were aligned to the Hong Kong catfish genome using TblastN [65]. Augustus (v3.3) [66] and Genscan (v3.0.4) [67] were used for de novo annotation. TopHat [68] was used to match the transcriptome data to the Hong Kong catfish genome. MAKER (v3.00) [69] was used to integrate the gene sets obtained by various methods into a non-redundant gene set. Annotation information for non-coding regions and variable shears was added using PASA. Finally, the proteins in the gene set were functionally annotated with protein databases (SwissProt, TrEMBL, KEGG, InterPro, GO, and NR).

tRNAscan-SE (v1.3.1) [70] was used to find tRNA sequences in the genome. BLASTN (v2.6.0) [71] was used to find rRNAs in the genome. In addition, INFERNAL [72] in Rfam (v14.1) was used to predict miRNA and snRNA sequences in the genome.

Genome evolution analysis

Gene family clustering was performed based on protein sequences of Hong Kong catfish and others 15 fish species, including L. chalumnae, L. oculatus, D. rerio, E. electricus, A. mexicanus, C. magur, S. meridionalis, P. hypophthalmus, I. punctatus, P. fulvidraco, E. lucius, G. morhua, O. latipes, O. niloticus, and T. rubripes. The protein sequences for each species were clustered based on sequence similarity using OrthoMCL (v14-137) [73]. Phylogenetic trees were constructed based on the shared single-copy direct homologous genes obtained by gene family clustering. Phylogenetic trees were constructed by the maximum likelihood method using RAxML (v8.2.11) [74]. The divergence time was predicted using MCMCtree in PAML (v4.9e) [75] and TimeTree for calibration. Using CAFÉ [76], random birth and death processes were used to simulate gene family expansion and contraction events for each lineage on the phylogenetic tree. The chromosomal genomes in Hong Kong catfish, I. punctatus, and P. hypophthalmus were included in a collinearity analysis using Mummer (v4.0.0beta2) [77].

Identification of the sex-determination region and potential sex-related genes

In a previous study, a high-density linkage map of Hong Kong catfish with 6453 SNP markers was constructed, and a sex-linked QTL was identified with a total map distance of 142.38 cM and 225 SNP markers [5]. SNP markers from the genetic linkage map were mapped to the chromosome-level genome using bwa (version 0.7.17) to determine the correspondence between the linkage groups (LGs) of the genetic linkage map and the chromosomes (chrs) of the genome assembly. The genomic region corresponding to the sex-linked QTL was obtained based on a comparison of the 225 SNP markers. The number of genes in the sex-linked QTL and annotation information were obtained. Gene functions were obtained using Swiss-Prot, PubMed, and NCBI databases.