The chromosome-level genome assembly of the red swamp crayfish Procambarus clarkii

Abstract

Red swamp crayfish, Procambarus clarkii, is the most cultured freshwater crayfish species. It attracts significant research attention due to its considerable economic importance. However, the limited availability of genome information has impeded further genetic studies and breeding programs. By utilizing Illumina, PacBio, and Hi-C sequencing technologies, we present a more comprehensive and continuous chromosome-level assembly for P. clarkii than the published one. The final genome size is 4.03 Gb, consisting of 2,358 scaffolds with a N50 of 42.87 Mb. Notably, 3.68 Gb, corresponding to 91.42% of the genome, was anchored to 94 chromosomes. The assembly comprises 70.64% repetitive sequences, including 5.21% tandem repeats and 65.40% transposable elements. Additionally, a total of 4,456 non-coding RNAs and 28,852 protein-coding genes were predicted in the P. clarkii genome, with 96.26% of the genes were annotated. This high-quality genome assembly not only represents a significant improvement for the genome of P. clarkii and provides insights into the unique genome evolution, but also offers valuable information for developing freshwater aquaculture and accelerating genetic breeding.

Background & Summary

Freshwater crayfish, comprising over 640 species, are crucial components of aquatic ecosystems, with Procambarus clarkii (Girard 1852) standing out as a representative species of significant economic value¹. In 2019, the global production of P. clarkii reached 2.17 million tons, valued at over US$14 billion². China dominates production with a staggering 95%, while the United States follows as the second-largest producer, contributing only 4% to the global production.

Native to the southern United States and northern Mexico, P. clarkii was introduced into China in 1929 and successfully established populations in more than 40 countries across the five continents, making it the most widely distributed freshwater crayfish globally^3,4. Moreover, P. clarkii demonstrates remarkable adaptability in various freshwater ecosystems, including rivers, lakes, ponds, and, notably, rice fields^5,6. In China, the cultivation area of P. clarkii has rapidly increased to 29.50 million mu in 2023, with the rice-crayfish integrated farming model accounting for 85.76%.

Despite being the most cultured crayfish species, the limited genetic information of P. clarkii has hindered further genetic studies and breeding programs. To address this, a chromosome-scale genome of P. clarkii was conducted using Illumina short-reads, PacBio long-reads, and Hi-C sequencing data (Fig. 1). A contig-level genome, consisting of 15,724 contigs with an N50 of 488.66 kb, represents a substantial improvement over the published genome⁷, which had an N50 of 216.75 kb. After Hi-C scaffolding, 2,358 scaffolds were obtained with an N50 of 42.87 Mb and a total length of 4.03 Gb. The largest 94 scaffolds were anchored to 94 chromosomes, spanning 3.68 Gb and corresponding to 91.42% of the assembled sequences. This genome includes 70.64% repeat sequences, 4,456 noncoding RNAs, and 28,852 genes, of which 96.26% were functionally annotated. The high-quality genome promises valuable insights into the molecular basis of P. clarkii, enriching crustacean genomic resources and providing foundational data for genome-wide selective breeding initiatives.

Fig. 1.

The pipelines overview of P. clarkii chromosome-level genome assembly and annotation.

Methods

Sample collection

Specimens of P. clarkii were obtained from a research base situated in Wuhan, Hubei, China (30°24′21″ N, 113°46′33″ E), where the rice-crayfish farming model was implemented. A total of twelve tissues were collected: ovary, hepatopancreas, gill, and muscle from a female crayfish, as well as brain, eyestalk, epidermis, stomach, heart, intestine, antennal gland, and testis from a male crayfish. All samples were immediately stored at −80 °C. Genomic DNA of muscle was extracted and subsequently used for Illumina, PacBio, and Hi-C sequencing, while total RNA of all twelve tissues was extracted for subsequent Iso-Seq and RNA-Seq. Both DNA and RNA were stored at −80 °C prior to sequencing.

Illumina sequencing and genome survey

The Illumina HiSeq X-ten instrument was used to sequence two libraries (270 bp and 500 bp), obtaining 576.87 Gb of raw data (Table 1). After filtering using SOAPnuke⁸ (v1.5.3) with parameter settings “-l 10 -q 0.4 -n 0.01”, resulting in 467.83 Gb of clean data. The clean data corresponded to approximately 117-fold haploid genome coverage. Of these, 120 Gb data were performed a genome survey analysis using a 21-mer frequency distribution with Jellyfish⁹ (v1.1.12) and GenomeScope¹⁰ (v2.0). This result estimated that the genome of P. clarkii is approximately 3.91 Gb in size, the heterozygosity rate is 1.01%, and the repeat content is 75.7% (Fig. 2a).

Table 1.

Statistics of sequencing data.

Data type	Instrument/Platform	Raw Data (Gb)	Clean Data (Gb)	N50 (bp)	GC (%)
WGS short reads	Illumina Hiseq X-ten	576.87	467.83	—	45.23
PacBio long reads	PacBio Sequel	431.89	—	19,710	42.22
Hi-C reads	MGISEQ-2000	847.06	835.58	—	43.65
Iso-seq subreads	PacBio Sequel	31.09	—	3,424	43.16
Transcriptome short reads	MGISEQ-2000	136.46	109.23	—	38.57

Fig. 2.

Genome assembly of P. clarkii. (a) The 21-mer analysis of the genome. len, estimated genome size in bp; uniq, unique percent; het, heterozygosity rate; kcov, kmer coverage; err, error rate; dup, duplication rate; k, kmer. (b) Hi-C assembly of chromosome interactive heatmap. Numbers on the axes show the chromosome length in Gb. The numbers at the bottom of the figure represents the chromosomes number. The ends of each link map to assembled scaffolds, with red dots represented the intersections of their corresponding loci in contigs. (c) Genomic features in a sliding window of 1 Mb. From the outermost to the innermost circle: i, chromosome ideograms; ii, protein-coding gene density; iii, TE content density; iv, LINE/CR1 content density; v, LTR/Gypsy content density; vi, GC content density.

PacBio sequencing and genome assembly

Twenty-four SMRTbell libraries were constructed with 20-kb DNA inserts, and subsequently sequenced using the PacBio Sequel instrument. PacBio sequencing generated 431.89 Gb of raw data with an N50 of 19.71 kb (Table 1). Subsequently, Falcon¹¹ (v0.2.2) and SMARTdenovo¹² (v1.0.0) were used to assemble the genome of P. clarkii, obtaining three assemblies: FALCON_1, FALCON_2, and SMARTdenovo versions (Table 2). The FALCON_1 and SMARTdenovo versions were assembled with default parameters, while the FALCON_2 version was assembled with following parameters “pa_DBsplit_option: -s600 -x2000; pa_HPCdaligner_option: -v -M24 -D200 -h600 -e.75 -s1000 -l4500 -k18; overlap_filtering_setting:–max_diff 90–max_cov 180–min_cov 10–bestn 10–n_core 16–min_len 10000”.

Table 2.

Different genome assemblies of P. clarkia.

	FALCON_1	FALCON_2	SMARTdenovo
N50 (Number)	238,067 bp (6,978)	342,212 bp (4,480)	319,513 bp (2,580)
N60 (Number)	169,238 bp (10,281)	256,456 bp (6,458)	231,795 bp (3,964)
N70 (Number)	108,440 bp (15,155)	184,151 bp (9,147)	164,297 bp (5,898)
N80 (Number)	59,535 bp (23,429)	120,262 bp (13,057)	111,317 bp (8,677)
N90 (Number)	36,170 bp (38,034)	65,344 bp (19,721)	66,461 bp (13,018)
Contig Number	72,012	34,869	22,275
Max Contig	2,482,788 bp	5,857,453 bp	7,134,225 bp
Total	6,634,750,326 bp	5,839,445,558 bp	3,746,471,481 bp

Comparing the total length of the three assemblies with the estimated genome size, we selected the FALCON_2 assembly for further genome polishing with FinisherSC¹³ (v1.22) and Pilon¹⁴ (v1.22). Then, the redundant contigs which shared > = 60% non-repetitive-region 17-mers (depth <100) with other sequences were filtered using the TrimDup¹⁵ tool (https://github.com/gigascience/rabbit-genome-assembler). After removing redundancies, we obtained a 4.02 Gb genome assembly, which is consist of 15,724 contigs with a N50 of 488.66 kb (Table 3).

Table 3.

Optimization of the P. clarkii genome assembly.

	FALCON_2	Polish	De-redundancy	Hi-C
N50 (Number)	342,212 bp (4,480)	343,380 bp (4,480)	488,656 bp (2,377)	42,870,550 bp (40)
N60 (Number)	256,456 bp (6,458)	257,297 bp (6,458)	386,822 bp (3,305)	36,398,550 bp (51)
N70 (Number)	184,151 bp (9,147)	184,825 bp (9,145)	294,505 bp (4,495)	34,570,824 bp (62)
N80 (Number)	120,262 bp (13,057)	120,720 bp (13,053)	212,348 bp (6,097)	29,397,883 bp (75)
N90 (Number)	65,344 bp (19,721)	65,466 bp (19,715)	127,446 bp (8,509)	19,119,945 bp (91)
Contig/Scaffold Number	34,869	34,869	15,724	2,358
Max Contig	5,857,453 bp	5,872,046 bp	5,872,046 bp	65,481,133 bp
Total	5,839,445,558 bp	5,858,056,529 bp	4,022,935,872 bp	4,029,667,872 bp

Hi-C sequencing and scaffolding

After sequencing on the BGI-Shenzhen MGISEQ-2000 platform, 847.06 Gb of raw Hi-C reads were obtained and then filtered, resulting in 835.58 Gb filtered data (Table 1). Bowtie2¹⁶ (v2.2.5) was used to align the Hi-C data to the FALCON_2 assembly with the parameters “--very-sensitive -L 30 --score-min L, -0.6, -0.2 --end-to-end –reorder”, obtaining 766,705,231 unique mapped read pairs. Of these, the valid interaction pairs identified using HiC-Pro¹⁷ (v2.5.0) achieved an impressive success rate of 83.61%. Subsequently, we used Juicer¹⁸ (v1.6.2) and 3D-dna¹⁹ (v180922) to anchor the contigs into chromosome-length scaffolds. The chromosome-level assembly contains 4.03 Gb in 2,358 scaffolds, with a N50 of 42.87 Mb (Table 3). Total length of 94 chromosomes was 3.68 Gb, with lengths range from 15.93 Mb to 65.48 Mb (Fig. 2b).

Repeat and noncoding RNAs annotation

Tandem repeats were annotated by Tandem Repeats Finder²⁰ (TRF, v4.09), while transposable elements (TEs) were annotated using de novo and homology-based predictions. We used RepeatMasker to annotate TEs in a de novo library constructed by RepeatModeler²¹ (v2.0.2). Meanwhile, we used RepeatMasker (v4.0.7) and RepeatProteinMask (v3.30) to identify the TEs with the Repbase transposable element library²². Using TEclass²³ (v2.1.3), the unknown TEs predicted above were further classified to four categories: DNA transposons, long terminal repeats (LTRs), long interspersed nuclear elements (LINEs), and short interspersed nuclear elements (SINEs). To sum, 70.64% repeats were identified in this genome, including 5.24% tandem repeats and 65.40% TEs (Table 4). This result is compared to the estimated repeat content of 75.7%.

Table 4.

Classification of repeat elements in the P. clarkii genome.

Types	Software	Repeat Size (bp)	% of genome
Tandem repeats	TRF	210,947,850	5.24
Transposable elements	RepeatMasker	741,299,344	18.43
	RepeatProteinMask	781,654,467	19.43
	RepeatModeler	2,481,286,644	61.68
	Combined TEs	2,630,818,761	65.40
Total	—	2,841,766,611	70.64

The tRNA was identified using tRNAscan-SE²⁴ (v1.3.138), while the miRNA was identified using the miRbase²⁵ (v22) database. Based on the Rfam²⁶ (v12.0) database, the rRNA was predicted by RNAmmer²⁷ (v1.2), while the snRNA was predicted with Infernal²⁸ (v1.1). A total of 837 tRNAs, 28 miRNAs, 2,819 rRNAs, and 772 snRNA were predicted (Table 5).

Table 5.

ncRNA statistics of P. clarkii genome.

Type	Number	Total length (bp)	Average length (bp)	% of genome
tRNA	837	62,940	75.19713	0.16
miRNA	28	2,145	76.60714	0.01
rRNA	2,819	304,302	107.9468	0.76
snRNA	772	132,946	172.2098	0.33

Gene prediction and function annotation

In this study, we performed gene prediction in a repeat-masked genome through transcriptome-based, homology-based, and de novo prediction methods. For transcriptome-based prediction, one PacBio Iso-seq library from muscle and eleven Illumina RNA-seq libraries from other tissues (including brain, eyestalk, gill, epidermis, stomach, heart, hepatopancreas, intestine, antennal gland, testis, and ovary) were constructed. Sequencing yielded 31.09 Gb of Iso-Seq data and 68.23 Gb of RNA-seq filtered data (Table 1). These data were aligned to the P. clarkii genome using Hisat2²⁹ (v2.1.0) with the parameters “–sensitive–no-mixed–no-discordant -X 1000 -I 1”, followed by assembly using StringTie³⁰ (v1.0.4). For homology-based prediction, BLASTN³¹ (v2.2.26) was used to align protein sequences from D. melanogaster, H. azteca, and L. vannamei against the P. clarkii genome, as well as GeneWise³² (v2.4.1) was then performed to compare the aligned proteins with the genome sequences to extract exon-intron information. Meanwhile, we performed de novo assembled for Iso-Seq and RNA-Seq data using Trinity³³ (v2.5.1) and predicted the gene structures with TransDecoder³⁴ (v5.5.0). De novo gene prediction was carried out using AUGUSTUS³⁵ (v3.4.0). The results from the homology-based, transcriptome-based, and de novo predictions were integrated using EvidenceModeler³⁶ (v5.0.2). A total of 28,852 genes were successfully predicted in the P. clarkii genome, with an average exon number of 4.5. The average lengths of gene, coding sequence (CDS), exon, and intron were 17,176 bp, 1,217 bp, 270 bp, and 3,833 bp, respectively (Fig. 2c, Table 6).

Table 6.

Summary statistics of predicted protein-coding genes of P. clarkii.

Species	Total number of genes	Average gene length (bp)	Average CDS length (bp)	Average exon number	Average exon length	Average intron length
P. clarkii	28,852	17,176.21	1,217.33	4.50	270.43	3,833.82

For functional annotation, six databases were used: NCBI non-redundant proteins (NR), SwissProt, TrEMBL, InterPro, Gene Ontology (GO), and KEGG. A total of 27,774 genes (96.26%) were successfully annotated with at least one functional protein database (Table 7). Among these, 11,223 genes were annotated in six databases, 6,734 in five databases, 4,141 in four databases, 3,121 in three databases, 1,758 in two databases, and 797 in only one database (Fig. 3).

Table 7.

Statistical results of gene function annotation.

Database	NR	SwissProt	TrEMBL	InterPro	GO	KEGG	Total
Number	26,774	19,662	27,568	20,858	13,121	23,265	27,774
Percentage	92.80%	68.15%	95.55%	72.29%	45.48%	68.15%	96.26%

Fig. 3.

Statistical results of gene annotation in the P. clarkii genome.

Data Records

All the raw sequencing data, including Illumina WGS, PacBio WGS, Hi-C WGS, Illumina RNA-Seq, and PacBio Iso-seq, have been submitted to the Sequence Read Archive (SRA) of National Center for Biotechnology Information (NCBI) under BioProject accession number PRJNA1123252, with accession number SRP516244³⁷. The genome assembly has been deposited at GenBank under the accession JBFOCG000000000³⁸. The annotation file was deposited at the Figshare repository³⁹.

Technical Validation

We assessed the quality of P. clarkii genome assembly in the four aspects. Firstly, contig N50 has achieved more than double the improvement to 488.66 Kb, which is much higher than published assembly (GCA_020424385.2) of P. clarkii or other closely related species^7,40–50. Secondly, we aligned the Illumina sequencing data back to the genome using BWA⁵¹ (v0.7.15), revealing a high mapping rate of 98.92%. Thirdly, we obtained 94.6% completed and 2.3% fragmented BUSCOs⁵² (v4.0) using the arthropoda_odb9 database. Fourthly, we analyzed the reasons for the differences in genome size of the two assemblies. Compared to the assessed genome size (3.07 Gb) of the published genome, our estimated size (3.91 Gb) is more consistent with the flow cytometry results (2 C = 8.4 ± 0.45 pg)^53,54. We also assessed collinearity with the previous published genome using MUMMER⁵⁵ (v3.0) with the parameters “–mum–mincluster 500”. As a result, 2.63 Gb of sequences of our assembly and 2.29 Gb of published genomes showed 1-to-1 Mummer-mapped alignments, while the other sequences with a high level of repeat content (77.76% and 75.78%) were classified as Mummer-unmapped, including many-to-many alignments and mismatches (Table 8). In addition, 135 Gb of our Illumina reads were aligned to the two genomes using BWA. The ratio of mapped reads to the two genomes was similar, but the mapped ratio of paired reads to our assembly (88.29%) is significantly higher than that to published genomes (79.38%) (Table 9). A total of 62.88 Mb and 148.29 Mb regions of the tow genomes have the coverage depth >100, and the accumulated mapped bases reach 17.40 Gb and 44.58 Gb, respectively. The theory genome size of these high-depth regions was estimated by accumulated mapped bases/Coverage depth (33X), resulting in 527.27 Mb and 1.32 Gb, respectively (Table 9). These results imply that our assembly has the higher integrity in high repeat regions.

Table 8.

The statistics of collinearity between the two P. clarkii genomes.

		This genome	Published genome (GCA_020424385.2)
Assembly Size		4,029,667,872 bp	2,735,413,669 bp
Mummer-mapped (1-to-1)	Size	2,629,236,616 bp	2,293,548,872 bp
	Repeat	1,590,050,739 bp	1,336,647,316 bp
	Repeat ratio	60.48%	58.28%
	Gene Number	23,823
Mummer-unmapped (many-to-many matches, or mismatch)	Size	1,400,431,256 bp	441,864,797 bp
	Repeat	1,089,010,611 bp	334,852,967 bp
	Repeat ratio	77.76%	75.78%
	Gene Number	5,029

Table 9.

The statistics of collinearity between the two P. clarkii genomes.

	This genome	Published genome
Total reads number	906,993,191
Total bases	135,158,456,739 bp
BWA mapped reads number	896,506,599	892,692,378
BWA mapped reads ratio	98.84%	97.97%
Paired reads number	800,760,657	723,269,574
Paired reads ratio	88.29%	79.38%
Unpaired reads number	95,745,942	169,422,804
Unpaired reads ratio	10.55%	18.59%
Total bases of high-depth region (Coverage Depth > 100)	62,884,502 bp	148,290,844 bp
Accumulated mapped bases	17,399,429,682 bp	44,581,751,584 bp
Theory genome size	527,255,445 bp	1,350,962,169 bp

Author contributions

Z.G., initiated, managed, and drove the genome project. R.H. and M.L. conceived the study. M.L. and Z.X. collected the animal material. M.X. and L.Q. managed data and provided bioinformatic support. M.L. and M.X. analyzed the data, with input from Z.G., R.H., J.W. and S.Z., M.L. carried out the experiments with Y.L. and X.L., M.L. and R.H. wrote the manuscript with contributions from Z.X., C.B., J.W., S.Z. and Z.G. All authors read and approved the final manuscript.

Code availability

All software and tools were used following the parameters mentioned in the Methods section or with default parameters.

Competing interests

The authors declare no competing interests.