Chromosome-level genome assembly of cotton thrips Thrips tabaci (Thysanoptera: Thripidae)

Abstract

Cotton thrip, Thrips tabaci is a major polyphagous pest widely distributed on a variety of crops around the world, causing huge economic losses to agricultural production. Due to its biological and genomic characteristics, this pest can reproduce quickly and develop resistance to various pesticides in a very short time. However, the lack of high-quality reference genomes has hindered deeper gene function exploration and slows down the development of new management strategies. Here, we assembled a high-quality genome of T. tabaci at the chromosome level for the first time by using Illumina, PacBio long reads, and Hi-C technologies. The 329.59 Mb genome was obtained from 320 contigs, with a contig N50 of 1.53 Mb, and 94.21% of the assembly was anchored to 18 chromosomes. In total, 17,816 protein-coding genes were annotated, and 96.78% of BUSCO genes were fully represented. In conclusion, this high-quality genome provides a valuable genetic basis for our understanding of the biology of T. tabaci and contributes to the development of management strategies for cotton thrip.

Background & Summary

Cotton thrip Thrips tabaci (Thysanoptera: Thripidae) is a polyphagous and devastating global insect pest species on numerous agricultural crops^1–4 (Fig. 1). T. tabaci is widespread in more than 120 countries and regions, including Asia, Europe, North America, South America, Australia, and Africa, and causes severe damage to more than 50 crops, including cotton, onions, leeks, garlic, cabbage, cucumbers, peas, strawberries, peppers and potatoes^1,5–7. The first and second instar larvae and adults of T. tabaci can feed on different plant organs to directly damage the crops or indirectly transmit plant orthotospoviruses, namely, Tomato spotted wilt virus (TSWV), Iris yellow spot virus (IYSV), Tomato yellow ring virus (TYRV) and Alstroemeria yellow spot virus (AYSV)^2,8–11. Pesticides application is the most widely used management strategy to control T. tabaci at present. However, T. tabaci still causes significant damage to the global agricultural system due to their small size, cryptic behaviour, polyophagy, short generation time, high reproductive capacity, dispersal to neighbouring farmfields or greenhouses, and transport along the international trade in agricultural products, while resistance to serial pesticides has been observed in several T. tabaci populations around the world, resulting in particular damage in turn^12–16. Therefore, to facilitate more innovative management strategies for this destructive pest, a deeper understanding of its genetics is needed and remains to be completed.

Fig. 1.

Morphological characteristics of Thrips tabaci across different development stages.

Previous studies have shown that polyphagous insects adapt to changing environments by inducing changes in gene expression related to detoxification enzymes when feeding on different crops or pesticides^17–19. The variation and selective expression of these genes make T. tabaci resistant to pesticides, which increases the difficulty of control^18,20,21. Therefore, there is an urgent need for genomic resources for molecular biology and physiology studies of pesticide resistance and reproduction in T. tabaci. Thysanoptera is one of the most important insects in the world, but only 6 thrips have had their genomes sequenced to date, namely Frankliniella intonsa, Frankliniella occidentalis, Megalurothrips usitatus, Stenchaetothrips biformis, Thrips palmi, and Aptinothrips rufus, while T. tabaci lacked the vital data of genome^22–24. In order to facilitate future research on the genetics, biology and ecology of T. tabaci, filling this knowledge gap will help provide theoretical support for optimizing management strategies for T. tabaci, which has important implications for pest control efforts.

In the present study, we propose a high-quality genome-assembly at chromosome level and conduct a whole life cycle transcriptome of T. tabaci using a combination of Illumina short-read sequencing, PacBio high fidelity (HiFi) reads, and high resolution chromosome conformation capture (Hi-C) techniques (Table 1). A 329.59 Mb genome was obtained from 320 contigs, with a contig N50 of 1.53 Mb (Tables 2), and 94.21% of the assembly (310.51 Mb of 329.59 Mb) was anchored to 18 chromosomes (Table 3) with a scaffold N50 of 16.56 Mb. We also predicted that the transposable elements accounted for 25.64% (12.40% retroelement and 13.24% DNA transposon) of the total genome sequence, and 9.22% of the total genome sequence was tandem repeats (Table 4). Besides, Eventually, 17,816 protein-coding genes (Table 5) was obtained, and the predicted genes were annotated and analyzed in eight databases: NR, EggNOG, GO, KEGG, TrEMBL, KOG, Swiss-Prot and Pfam. 13, 569 genes were annotated in the GO database, 12, 983 genes were annotated in the KEGG database, 10, 813 genes were annotated in the KOG database, 14, 275 genes were annotated in the Pfam database, 10, 404 genes were annotated in the Swiss-Prot database, 15, 652 genes were annotated in the TrEMBL database, 12, 768 genes were annotated in the EggNOG database, and 15, 114 genes were annotated in the NR database. Finally, A total of 16, 209 genes were annotated in all databases, accounting for 90.98% of all protein-coding genes (Table 6). Our genomic features of T. tabaci will lay a foundation for future ecological studies of thrips and provide a genetic basis for further studies of this polyphagous pest.

Table 1.

Statistics of sequencing data of Thrips tabaci genome.

Library type	Usage	Insert Size (bp)	Clean Data (Gb)	Coverage (X)
Illumina	Genome survey	350	24.51	87.99
PacBio	Genome assembly	20000	25.59	94.70
Hi-C	Hi-C assembly	150	56.30	208.34
RNA-Seq (Illumina)	Anno-evidence	150	85.91
RNA-Seq (PacBio)	Anno-evidence	20000	33.07

Table 2.

Statistics of genome assembly of Thrips tabaci at the chromosomal-level.

Features	Values
Estimate the genome size (Mb)	278.54
Total length (Mb)	329.59
Longest scaffold length (bp)	21,765,751
Contig numbers	320
Contig N50 (bp)	1,531,534
Reads length mean (bp)	16,193
Scaffold N50 (bp)	16,564,241
Scaffold N90 (bp)	14,161,603
GC (%)	53.28
Anchored to chromosome (Mb, %)	310.51 (94.21%)

Table 3.

Statistics of Hi-C assembly results.

Pseudo-chromosomes	No. Cluster	Cluster Length (bp)	No. Order	Order Length (bp)
Chr01	25	23,590,669	23	21,763,551
Chr02	21	23,320,197	19	21,720,082
Chr03	24	23,517,496	21	21,358,915
Chr04	20	21,294,565	19	19,667,489
Chr05	21	20,467,540	20	18,821,432
Chr06	9	18,215,419	9	18,215,419
Chr07	20	19,122,245	18	17,483,537
Chr08	9	17,007,937	9	17,007,937
Chr09	20	16,562,341	20	16,562,341
Chr10	18	17,444,087	16	16,544,672
Chr11	20	19,763,899	18	16,343,346
Chr12	21	18,427,373	17	16,351,134
Chr13	16	16,425,660	15	15,849,037
Chr14	17	15,711,168	16	15,406,219
Chr15	18	15,102,040	18	15,102,040
Chr16	17	15,049,817	16	14,744,272
Chr17	12	14,856,180	11	14,160,603
Chr18	12	13,706,612	11	13,410,671
Total (Ratio %)	320 (100.00)	329,585,245 (100.00)	296 (92.50)	310,512,697 (94.21)

Table 4.

Classification of repeat elements in Thrips tabaci genome.

Repeat types		Number	Length (bp)	Percent (%)
Retroelement	DIRS	1	73	0	12.40
	LINE	20,517	4,663,532	1.41
	SINE	1,809	223,456	0.07
	LTR/Copia	6,199	3,260,579	0.99
	LTR/ERV	1,533	210,523	0.06
	LTR/Gypsy	31,415	14,185,724	4.30
	LTR/Ngaro	334	26,769	0.01
	LTR/Pao	195	15,106	0
	LTR/Unknown	118,446	18,271,408	5.54
DNA transposon	Academ	102	60,099	0.02	13.24
	CACTA	1,044	89,800	0.03
	Crypton	249	47,534	0.01
	Dada	21	1,156	0
	Ginger	117	6,997	0
	Helitron	120,799	17,559,717	5.33
	IS3EU	38	3,155	0
	Kolobok	134	16,700	0.01
	Maverick	482	229,551	0.07
	Merlin	3	152	0
	Mutator	10	1,274	0
	P	2	45	0
	PIF-Harbinger	80	30,609	0.01
	PiggyBac	87	32,058	0.01
	Sola	17	2,881	0
	Tc1-Mariner	159	12,005	0
	Unknown	99,195	25,342,075	7.69
	Zator	41	18,447	0.01
	Zisupton	61	3,487	0
	hAT	1,375	174,669	0.05
Unknown		42	8,351	0
srpRNA		5	1,455	0
Tandem repeat	Microsatellite (1–9 bp units)	175,172	4,548,089	1.38	9.22
	Minisatellite (10–99 bp units)	119,995	9,614,460	2.92
	Satellite (> = 100 bp units)	11,590	16,241,158	4.93
Total		711,269	114,903,094	34.86

Table 5.

Gene annotation statistics of Thrips tabaci genome.

Features	Results
Number of annotated genes	17,816
Total lineage BUSCOs	1367
Complete BUSCOs and Ratio	1328 (97.15%)
Complete and single-copy BUSCOs and Ratio	1032 (75.49%)
Complete and duplicated BUSCOs and Ratio	296 (21.65%)
Fragmented BUSCOs and Ratio	1 (0.07%)
Missing BUSCOs and Ratio	38 (2.78%)
Average gene length (bp)	7837.37
Number of Exon	138,164
Average Exon length (bp)	1760.30
Average Exon count per gene	7.76
Number of CDS	138,153
Average CDS length (bp)	1758.72
Average CDS per gene	7.75
Number of Intron	120,348
Average Intron length (bp)	6077.08
Average Intron per gene	6.76

Table 6.

Functional annotation statistics of Thrips tabaci genome.

Annotation type	Genes number	Percent (%)	Homepage
GO	13,569	76.16	http://www.geneontology.org/
KEGG	12,983	72.87	http://www.genome.jp/kegg/
KOG	10,813	60.69	http://www.ncbi.nlm.nih.gov/KOG/
Pfam	14,275	80.12	http://pfam.xfam.org/
Swiss-Prot	10,404	58.40	http://www.uniprot.org/
TrEMBL	15,652	87.85	http://www.uniprot.org/
EggNOG	12,768	71.67	http://eggnog5.embl.de/#/app/home
NR	15,114	84.83	ftp://ftp.ncbi.nih.gov/blast/db/
Total annotated genes	16,209	90.98

Methods

Sample preparation and genomic DNA sequencing

A colony of Thrips tabaci originally collected from cotton field in Henan Province of China, was reared in the laboratory for approximately 100 generations. Adults were fed on Brassica oleracea and kept at controlled conditions of 25 ± 0.5 °C, 60 ± 5% relative humidity, and a photoperiod of 16 h Light: 8 h Dark²⁵. Briefly, pupal thrips were decontaminated by immersing in 1% sodium hypochlorite solution for 5 min, followed by rinsing in sterile water and immersion in 70% ethanol twice, and then rinsing in sterile water again. Before genomic DNA and RNA extraction, samples were rapidly transferred to collection tubes and flash frozen in liquid nitrogen, then stored at −80 °C.

We prepared approximately 4,000 pupae of T. tabaci for genome sequencing. Genomic DNA was extracted using the QIAGEN^® Genomic kit (QIAGEN, Dusseldorf, Germany) according to the manufacturer’s instructions. The purity and concentration of genomic DNA was determined by 0.75% agarose gel electrophoresis, Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, USA), and Qubit TM3 Fluorometer (Invitrogen, USA), successively.

To construct the library, the total genomic DNA was randomly sheared into fragments of ~15 kb. The SMRTbell library was constructed using the SMRTbell Express Template Prep kit 2.0 (Pacific Biosciences). Before annotating the data obtained by PacBio readings, we performed a series of preprocessing procedures. The sheared 10 μg of DNA was brought into the first enzymatic reaction to remove the single-strand dangling, and then treated with repair enzymes to repair any damage that may exist on the DNA backbone. After DNA damage repair, ends of the double-stranded fragments were polished and subsequently tailed with an A-overhang at the 3′end. Ligation with T-overhang SMRTbell adapters was performed at 20 °C for 60 minutes. Following ligation, the SMRTbell library was digested by exonuclease and purified with 0.45X AMPure PB beads. After library characterization, the Sage ELF system (Sage Science, Beverly, MA) was used to perform a size selection step on 3 μg to collect SMRTbells 15–18 kb. After size selection, the library was purified with 1X AMPure PB beads. The size and quantity of libraries were evaluated using FEMTO Pulse and Qubit dsDNA HS reagents Assay kits. The sequencing primers and Sequel II DNA polymerase were annealed and combined into the final SMRTbell library, respectively. The library was subjected to diffusion loading at 55 pM plate concentration.

The long-read library was sequenced on the PacBio Sequel II platform (Pacific Biosciences, USA) at BioMarker (Beijing) Co., Ltd., and circular consensus (CCS) reads were generated. The 150 bp paired-end short read sequencing libraries were sequenced on the Illumina NovaSeq 6000 platform. After filtering out the adapters and low-quality reads, approximately 25.59 Gb of subreads (coverage: 94.7 × ) were obtained from the PacBio long-read sequencing (Table 1). The PacBio reads had an average length of 16,193 bp, with an Contig N50 length of 1,531,534 bp (Table 2). The Illumina platform generated a total of 24.51 Gb (coverage: 87.99 × ) of clean data with an average insert size of 350 bp (Table 1).

Hi-C library preparation and sequencing

The Hi-C (high-throughput chromatin conformation capture) technique was used to construct the chromosome-level genome assembly of T. tabaci²⁶. For Hi-C library construction, fresh tissues from 2000 pupae individuals were to be used. Formaldehyde is used to fix the sample, cross-link intracellular proteins with DNA, DNA with DNA, preserve their interactions, and maintain the 3D structure in the cell. The DNA is digested with the restriction enzyme DpnII, resulting in sticky ends on both sides of the crosslink. Finally, the DNA was broken into fragments of 300 bp to 700 bp, and streptaviin magnetic beads were used to capture DNA fragments containing interacting relationships for library construction. After the library inspection is qualified, the Illumina platform was used for high-throughput sequencing, and the sequencing read length is paired-end 150 bp. The Hi-C library was constructed following the standard library preparation protocol, and 56.30 Gb of clean data was generated (Table 1).

Transcriptome sequencing

Transcriptome samples were prepared at all developmental stages of T. tabaci including the 1^st and 2^nd instar nymphs, pupae and adults of T. tabaci, respectively. Separate total RNA was extracted from larvae, pupae, and adult samples collected above, using the TRIzol reagent (Thermo Fisher Scientifc, Waltham, USA). The complementary DNA (cDNA) library was constructed and sequenced on an Illumina Novaseq 6000 platform. After library construction, Qubit2.0 and Agilent 2100 were used to determine the concentration and insert size of the library, and Q-PCR was used to accurately quantify the effective concentration of the library to ensure the library quality. A total of 85.91 Gb clean RNA-seq data was obtained after the following quality control: removal of reads containing connectors and removal of low-quality reads (including reads that remove more than 10% of N; the reads with mass value Q ≤ 10 accounted for more than 50% of the entire read) (Table 1).

In addition, the full-length cDNA of mRNA was synthetised using the SMARTer™ PCR cDNA Synthesis Kit (Pacifc Biosciences, USA) for library construction. Subsequently, full-length transcriptome sequencing was performed on the PacBio Sequel II platform, and circular consensus (CCS) reads were generated. A total of 33.07 Gb full-length transcriptome data were obtained after the following quality control: according to the adaptor in the sequence, all the original sequences were converted into CCS sequences, and the CCS sequences were polished to obtain the quality information of the sequences. According to whether there were 3′ primers, 5′ primers and PolyA in the CCS sequence, the sequence was divided into full-length sequence and non-full-length sequence. The full-length sequences from the same transcript were clustered, and the similar full-length sequences were clustered into a cluster. Each cluster obtained a consistent sequence and extracted high-quality sequences (Table 1). Ultimately these high-quality sequencing data were mapped to the assembled genome to identify gene transcript levels²⁷.

Estimation of genomic characteristics

Firstly, in order to evaluate the genomic characteristics of T. tabaci, the genome was investigated using next-generation sequencing, and 24.51 GB of high-quality data was generated, the short reads from the Illumina platform were quality filtered by Fastp (version 0.21.0)²⁸ using the parameters of ‘-q 10 -u 50 -y -g -Y 10 -e 20 -l 100 -b 150 -B 150’. The high-quality filtered reads were used for further genome size estimation. In order to determine whether the DNA of the extracted samples was contaminated, 10,000 single-ended reads were randomly selected from a 350 bp library sequenced and compared to the NT library by BLAST (ncbi-blast + , version 2.2.29)²⁹ with the parameter set to ‘-num_descriptions 100-num_alignments 100-evalue 1e-05’. The libraries sequenced by Illumina platform were compared with plastids for SOAP (version2.21)³⁰, and the parameter was set to ‘-m 260-x 440’ to evaluate the extranuclear DNA content in the libraries and ensure the integrity of the genome assembly. We counted the 19-kmers using Jellyfish (version 2.1.4)³¹ with the following parameters ‘-h 10000000000’ and calculated the genome features using Genomescope (version2.0)³² with the parameters of ‘-k 19 -p 6 -m 100000’. Through fitting various ploidy data, it was found that the Kmer distribution map had the best fitting degree when it was hexaploid (Fig. 2A). Using hexaploid as fitting standard, the kmer depth corresponding to the first peak is 13.5, and the length of a single genome is approximately 278.54 Mb (Table 2). According to the distribution of kmer, it is estimated that the repetitive sequence content was about 33.14%, the heterozygosity was about 2.47%, and the GC content of the genome was about 51.39% (Fig. 2A).

Fig. 2.

Genome assembly and temporal transcriptome of Thrips tabaci. (A) Genome scope profiles of 19-mer analysis. (B) Hi-C interactive heatmap of eighteen linkage pseudo-chromosomes in Thrips tabaci genome. Color indicates the intensity of the interaction signal. The darker the color, the higher the intensity. (C) Circle genome landscape of Thrips tabaci. Circle a represents chromosomes, while circles b-e indicate repeat density, SSR density, gene density, and GC content of each respective chromosome, respectively. (D) Protein-coding gene prediction of Thrips tabaci through three strategies.

De novo genome assembly

For the quality control of PacBio long reads (CCS) data, we mainly performed error correction of identifiable haplotypes. For CCS reads, although its accuracy is high, some errors are still retained. Hifiasm (version0.19)³³ will read all CCS reads into memory for all-vs-all comparison and error correction. Based on the overlap informations between reads, if a base on a read is different from other bases and supported by at least three reads, it is considered to be a SNP and retained, otherwise it is considered to be wrong and corrected. It is worth noting that Hifiasm (version0.19)³³ only uses the data of the same haplotype for error correction, thereby avoiding overcorrection and retaining heterozygous variation information from different haplotypes. In this step, Hifiasm (version0.19)³³ can phasing the heterozygous SNP.

High accuracy CCS data was used for genome assembly by the Hifiasm (version0.19)³³ with parameters ‘-2, -4’. For genomes with high heterozygosity, the initial assembly may assemble all the heterozygous fragments, resulting in a larger than expected genome. The purge_dups³⁴ was used to resolve the haplotigs and overlaps in a primary assembly based on read depth, while Pilon³⁵ and Racon³⁶ were used to polish the assembly. The scaffold pipeline of the genome was deconstructed in equal lengths of 50 Kb and reassembled using Hi-C technology²⁶. The locations that could not be restored to the original assembly sequence are listed as candidate error regions, and the locations with low Hi-C coverage depth in this region are identified as error points, thus completing the error correction of the original assembled genome. For anchored contigs, clean read pairs were generated from the Hi-C library and were mapped to the polished T. tabaci genome using BWA (version0.7.17)³⁷ with the default parameters. Paired reads with mapped reads to a different contig were used to do the Hi-C associated scaffolding. Self-circle ligation, non-ligation and other invalid reads, such as Dangling Ends, Re-ligation, and Dumped Pairs were filtered. HiC-Pro (version2.10.0)³⁸ can identify valid interaction pairs and invalid interaction pairs in Hi-C sequencing results by analyzing and comparing the results, and realize the quality evaluation of Hi-C library. We then successfully clustered 320 contigs into 18 groups (Table 3) using the agglomerative hierarchical clustering method in LACHESIS (version 2e27abb)³⁹. Furthermore, 296 of 320 contigs were successfully ordered and oriented with the length of 310,512,697 bp via LACHESIS as well (Table 3). Finally, we obtained the first high-quality assembled genome of T. tabaci at the chromosome level. The genome was consisted of 320 contigs with a total length of 329.59 Mb, which was similar to the predicted size of 278.54 Mb, and with a scaffold N50 of 16.56 Mb, maximum length of 21.77 Mb, and GC rate of 53.28%. The analysis of Hi-C data helped to anchor 296 (92.50%) contigs of 310.51 (94.21%) Mb sequence to 18 pseudo-chromosomes, which were well-distinguished from each other based on the chromatin interaction heatmap (Tables 2, 3; Fig. 2B).

Repetitive elements and noncoding RNA annotation

Transposon elements (TE) and tandem repeats were identified using a combination of homology-based and de novo approaches. We first customized a de novo repeat library of the genome using RepeatModeler (version2.0.1)⁴⁰ (http://www.repeatmasker.org/RepeatModeler/) with parameters ‘BuildDatabase -name & RepeatModeler -pa 12’, which can automatically execute two de novo repeat finding programs, including RECON (version1.0.8)⁴¹ and RepeatScout (version1.0.6)⁴². Then full-length long terminal repeat retrotransposons (fl-LTR-RTs) were identified using both LTR_FINDER (version2.8)⁴³ with parameters ‘-w 2 -C -D’ and LTRharvest (version1.5.9)⁴⁴ with default parameters. The high-quality intact fl-LTR-RTs and non-redundant LTR library were then produced by LTR_retriever (version2.9.0)⁴⁵ with default parameters. Non-redundant species-specific TE library was constructed by combining the de novo TE sequence library above with the well-known Dfam (version3.5)⁴⁶ database. Final TE sequences in the T. tabaci genome were identified and classified by homology search against the library using RepeatMasker (version4.12)⁴⁷ with parameters ‘-nolow -no_is -norna -engine wublast -parallel 8 -qq’. Tandem repeats were annotated by Tandem Repeats Finder (TRF) (version409)⁴⁸ with parameters ‘2 7 7 80 10 50 500 -d -h’ and MIcroSAtellite identification tool (MISA) (version2.1)⁴⁹ with default parameters. In total, 34.86% of the assembled genome was classified as repetitive sequences in the 329.59 Mb genome, including transposable elements (TEs) with a sequence length of 84,499,387 bp, accounting for 25.64% of the whole genome, and tandem repeats with a sequence length of 30,403,707 bp, accounting for 9.22% of the whole genome (Table 4).

Non-coding RNAs are RNAs that do not encode proteins, including microRNA, rRNA, tRNA and other RNAs with unknown functions. According to the structural characteristics of different non-coding RNAs, several specific strategies are used to predict corresponding non-coding RNAs. tRNAscan-SE (version1.3.1)⁵⁰ was used to identify tRNA with default parameters. Prediction of rRNA was mainly made using barrnap (version0.9)⁵¹ with parameters ‘kingdom euk–threads 1’. miRNA, snoRNA and snRNA were predicted based on the Rfam (version14.5)⁵² database via Infenal (version1.1)⁵³ with parameters ‘cpu 3–rfam’. Finally, a total of 2,980 tRNAs, 61 rRNAs, and 27 miRNAs were obtained (Table 7).

Table 7.

Statistics of noncoding RNA in Thrips tabaci genome.

No. tRNA	No. rRNA	No. miRNA
2980	61	27

Gene Prediction and Functional Annotation

Before the gene annotation of CCS reads, we enriched the mRNA containing polyA tail by primers with Oligo-dT, and then reverse transcribed the mRNA using Iso-Seq RT enzyme. By adding template switching oligo (Template Switch Oligo, TSO), the synthesized cDNA was amplified by PCR. Then, the full-length cDNA was repaired by damage repair, end repair, and end plus A tail. Finally, the SMRT dumbbell-shaped sequencing adaptor was connected, and the sequencing primers were combined to bind the DNA polymerase to form a complete SMRT-bell sequencing library by using Iso-Seq (version 4.0.0) with default parameters.

We integrated three approaches, namely, de novo prediction, homology search, and transcript-based assembly, to annotate protein-coding genes in the genome. Based on the genome sequence, we used Augustus (version3.2.3)⁵⁴ and SNAP (version2006-07-28)⁵⁵ for ab initio gene prediction with default parameters. For homo-based approaches, GeneModelMapper (GeMoMa) (version1.7)⁵⁶ was used with Aptinothrips rufus, Drosophila melanogaster, Frankliniella occidentalis, Trips palmi, and Megalurothrips usitatus (Table 8) as references with parameters ‘run.sh mmseqs’. For the transcript-based methods, RNA-seq reads were mapped to our assembled reference genome above by using Hisat (version2.1.0)⁵⁷ with the parameters ‘dta -p 10’. Stringtie (version2.1.4)⁵⁸ was then applied with parameters ‘p 2’ to assemble the mapped reads into transcripts. Genes were predicted from the assembled transcripts using GeneMarkS-T (version5.1)⁵⁹ with default parameters. Meanwhile, the PASA (version2.4.1)⁶⁰ was utilized to predict genes based on the unigenes assembled by Trinity (version2.11)⁶¹ with default parameters ‘genome_guided_bam’. Full-length transcripts from the PacBio sequencing were compared using gmap (version 2020-06-30) with parameters ‘cross-species–nthreads = 4 -f 2’, and then used PASA (version2.4.1)⁶⁰ for gene prediction. Finally, we merged predicted genes obtained from strategies of homology-based, de novo-derived, and transcripts, to generate the high-confidence gene set using the EVidenceModeler (versio1.1.1)⁶² with default parameters. In total, 17,816 protein-coding genes with an average length of 7837.37 bp were obtained in the assembled T. tabaci genome (Table 5). The average length of coding sequence (CDS) was 1758.72 bp with a total number of CDS of 138,153. The average exon length was 1760.30 bp and the average Exon number of each gene was 7.76. The insecta database in BUSCO contains 1,367 conserved core genes (Table 5). We used Benchmarking Universal Single-Copy Orthologs (BUSCO, version5.2.2)⁶³ with parameters ‘m prot’ to evaluate the completeness of gene prediction, where 97.15% of BUSCO genes were present in the predicted genes (Table 5), indicating a high completeness of gene prediction in T. tabaci.

Table 8.

The number of differential expressed genes in the pairwise comparisons of Thrips tabaci across 1^st instar nymphs (N1), 2^nd instar nymphs (N2), pupa and adult.

Groups	DEGs_total	DEGs_up	DEGs_down
N1_vs_N2	2944	1530	1414
N1_vs_Pupa	2397	1040	1357
N1_vs_Adult	6868	3005	3863
N2_vs_Pupa	1416	774	642
N2_vs_Adult	8603	3895	4708
Pupa_vs_Adult	6537	2674	3863

Gene structure and annotations were identified according to the best match of the alignments to the public databases including National Center for Biotechnology Information (NCBI) Non-Redundant (NR) (ftp://ftp.ncbi.nih.gov/blast/db/), EggNOG⁶⁴, KOG (http://www.ncbi.nlm.nih.gov/KOG/), GO (http://www.geneontology.org/), TrEMBL⁶⁵ and Swiss-Prot protein databases⁶⁵ (Table 6) using diamond (2.0.4.142) with parameters ‘masking 0 -e 0.001’. The data were also compared with the Kyoto Encyclopedia of Genes and Genomes (KEGG) database⁶⁶ (Table 6), with an E-value threshold of 1E-5. The motifs and domains within gene models were identified by Pfam databases⁶⁷ (Table 6). In total, approximately about 90.98% (16209 of 17816 total predicted genes) of the predicted protein-coding genes could be annotated in those databases above (Fig. 2D, Table 6). Finally, the number of EVM-integrated genes supported by the three prediction methods were counted separately, and 13,848 genes were predicted by all three methods (Fig. 2D).

Temporal transcriptome of T. tabaci across nymphal, pupal, adult stage

Samples were collected from all developmental stages of T. tabaci, namely, first instar nymph, second instar nymph, pupa and adult. Fifty T. tabaci individuals were placed in each 1.5 mL collection tube with 3 replicates per stage. TRIzol reagent was used for RNA extraction. Transcriptome sequences were obtained using the same procedure as described in the section of “Transcriptome sequencing” above.

Hisat (version2.1.0)⁵⁷ was used to quickly align the clean reads obtained from transcriptome sequencing accurately with the assembled genome of T. tabaci to obtain the location information of Reads on the reference genome. Then Stringtie (version2.1.4)⁵⁸ was used to assemble the above reads, and the transcripts were reconstructed for subsequent analysis with default parameters. Finally, FPKM (Fragments Per Kilobase of transcript per Million fragments mapped) was standardized as an indicator to measure the level of transcript or gene expression⁶⁸.

Based on the count value of genes in each sample, DESeq2⁶⁹ was used for differential expression gene screening and differential analysis. The False Discovery Rate (FDR) was obtained by correcting the significance p-value of the difference, in order to reduce false positives caused by changes in the expression of a large number of genes⁷⁰. Finally, |log₂FC| ≥ 2 and FDR < 0.01 were used as the criteria in the process of significantly differential expression gene detection. Hierarchical clustering analysis of all identified DEGs was performed, and genes with the same or similar expression patterns in different samples were clustered by hclust () function in R-packets. DEGs were mapped to GO terms and KEGG pathways, and an enrichment analysis was performed to identify any over-representation of GO terms and KEGG pathways by hypergeometric test (Figs. 3, 4; Table 8).

Fig. 3.

Differential expressed genes in pairwise comparison of Thrips tabaci across different developmental stages. (A) Venn diagrams of DEGs among those pairwise comparisons. (B–G) Volcanic maps of different expressed genes in separate comparison among different developmental stages. The red and green balls represent the significantly up- and down-regulated expressed genes, respectively.

Fig. 4.

Significantly enriched KEGG pathways of DEGs in these pairwise comparison of Thrips tabaci at different developmental stages.

Data Records

Genomic Illumina sequencing data were deposited in the Sequence Read Archive at NCBI under accession number SRR26448191⁷¹. Genomic PacBio sequencing data were deposited in the Sequence Read Archive at NCBI under accession number SRR26417911⁷². Hi-C sequencing data were deposited in the Sequence Read Archive at NCBI under accession number SRR26401083⁷³. RNA-seq data were deposited in the Sequence Read Archive at NCBI under accession number PRJNA1028763 and SRR26404855⁷⁴. The final assembled Thrips tabaci genome has been submitted to NCBI under accession number GCA_040581495.1⁷⁵. The annotation files of the Thrips tabaci genome have been deposited at figshare⁷⁶.

Technical Validation

DNA integrity

Nanodrop 2000 spectrophotometer (Thermo Fisher Scientifc, USA) and QubitTM3Flurometer (Thermo Fisher Scientific, USA) were used to detect the concentration of extracted DNA. Absorbance of obtained DNA at 260/280 nm and 260/230 nm were both about 1.8. The quality of genomic DNA was detected by agarose gel electrophoresis. The main band size of DNA fragments was ≥ 23 K, the degradation band was > 5 K. There was no contamination in the sample holes, which proved that good integrity of DNA molecules was observed in this study.

Assessment of genome assemblies

We assess the integrity and accuracy of the genome in four main ways: First, the short sequences obtained by Illumina sequencing platform were compared with the assembled genome through BWA (version0.7.17)³⁷, and the integrity of the assembled genome and the uniformity of sequencing coverage can be assessed by statistical comparison rate, proportion of covered genomes, and depth distribution. The results showed that 164,272,072 clean reads were obtained, of which 155,286,329 were located to the reference genome, accounting for 94.53% of all Clean Reads (Table 9). Secondly, Minimap2⁷⁷ was used to compare HiFi reads with assembled genomes. The results showed that 1,580,153 clean reads were obtained by three-generation sequencing via PacBio CCS technology, and 1,532,715 clean reads were located to the reference genome, accounting for 97.00% of all clean reads (Table 9). Third, the Core Eukaryotic Genes Mapping Approach (CEGMA) was used to evaluate the assembled genome by selecting conserved genes (458 genes) existing in eukaryotic model organisms to form a core gene library to assess the integrity of the assembled genome. The results showed that 452 genes were identified in the assembled genome, accounting for 98.69% of the total. Finally, the single-copy gene set constructed by BUSCO (version5.2.2)⁶³ was compared with the assembled genome, and the ratio and completeness of the comparison were evaluated. The results showed that the completeness of BUSCO evaluation was 96.78% with only 0.07% fragmented BUSCOs and 3.15% missing BUSCOs (Table 9). All of the above results indicate that our assembled Thrips tabaci genome has high integrity and accuracy. These BUSCO results were compared with the genome integrity of other thrips species, all of which were comparable, such as T. palmi (97.20%), M. usitatus (97.40%), and A. rufus (95.00%) (Table 10).

Table 9.

Assessment metrics for the final genome assembly of Thrips tabaci.

Items	Types	Ratios
Genome completeness	Complete BUSCOs (C)	96.78%
	Complete and single-copy BUSCOs (S)	87.93%
	Complete and duplicated BUSCOs (D)	8.85%
	Fragmented BUSCOs (F)	0.07%
	Missing BUSCOs (M)	3.15%
Genome accuracy	Mapping short-reads rate	94.53%
	Mapping HiFi reads rate	97.00%

Table 10.

Comparisons of genome assemblies of different thrips species.

Species	Assembly level	Genome size (Mb)	Pseudo-chromosomes number	Scaffold N50 (Kb)	BUSCO (%)	GC (%)	Data source
Thrips tabaci	Chromosome	329.59	18	16,564	96.78	53.28	In this study
Megalurothrips usitatus Ref1	Chromosome	238.14	16	13,852	97.40	55.90	10.1038/s41597-023-02164-5
Megalurothrips usitatus Ref2	Chromosome	247.82	16	14,859	98.60	55.40	10.3390/ijms241411268
Trips palmi	Chromosome	237.85	16	14,670	97.20	53.90	10.1111/1755-0998.13189
Stenchaetothrips biformis	Chromosome	338.86	18	18,207	96.60	51.09	10.1038/s42003-023-05187-1
Frankliniella occidentalis	Scaffold	274.99	15	4,180	98.50	48.40	10.1186/s12915-020-00862
Aptinothrips rufus	Contig	339.92		5	95.00	48.60	http://v2.insect-genome.com/Organism/87

The 3 C (contiguity, completeness and correctness) criterion: 1) Contiguity: a total of 24.51 Gb data were obtained by short reads sequencing. The total sequencing depth was about 87.99 X, the GC content was about 51.39%, and the genome size was about 278.54 Mb; The amount of data obtained by long reads sequencing was 25.59 Gb, the depth was about 94.7 X, the average length of reads was 16.19 k, the number of contigs/scaffolds was 320, the Contig N50 was 1.53 Mb, the GC content was about 53.28%, and the total length of the genome sequence was 329.59 Mb. 2) Completeness: In the long reads, the insecta database of OrthoDB 10 was selected. The number of genes in the core gene set were 1,367. The number and proportion of complete single-copy core genes in the core gene set were 1202 and 87.93%, respectively. The number and proportion of complete multi-copy core genes in the core gene set were 121 and 8.85%, respectively. The number and proportion of complete core genes in the core gene set (including single-copy and multi-copy) were 1323 and 96.78%. The sequencing data were compared with the assembly results to evaluate the data coverage. In short reads, the number of clean reads was 164,272,072, the number of clean reads mapped to the reference genome was 155,286,329, and the percentage of clean reads mapped to the reference genome to all clean reads was 94.53%. In long reads, the number of clean reads was 1,580,153, the number of clean reads mapped to the reference genome was 1,532,715, and the percentage of clean reads mapped to the reference genome to all clean reads was 97.00%. 3) Correctness: CEGMA (Core Eukaryotic Genes Mapping Approach) evaluation is to select conserved genes (458 genes) existing in eukaryotic model organisms to form a core gene library, and to evaluate the assembled genome with software such as tblastn to evaluate the correctness of the assembled genome. The number of conserved genes contained in the assembled genome was 452, accounting for 98.69% of the 458 conserved genes. The assembled genome contains 244 highly conserved genes, accounting for 98.39% of the 248 highly conserved genes. The short reads data were re-aligned and compared with the assembled genome. The alignment rate was 94.53%, the cover degree was 97.77%, and the average sequencing depth was 64 × ; The long reads data were remapped and compared with the assembled genome. The alignment rate was 97.00%, the cover degree was 99.99%, and the average sequencing depth was 64 × . Finally, we obtained a Thrips tabaci genome sequence with high contiguity, completeness and correctness.

Author contributions

J.J., J.C., J.L. and X.Z. conceived the project; Y.G., C.X., X.Z., L.W., K.Z. and D.L. performed the experiments; Y.G., X.W. and M.X. performed the bioinformatic analyses; Y.G., H.H. and L.C. evaluated the results; Y.G. and J.J. wrote the manuscript. All authors read and approved the final manuscript.

Code availability

All bioinformatics tools and software used in this study were acquired in public databases and executed in accordance with published bioinformatics tools manuals and protocols. The software version and parameters were described in the method without the use of specific code or scripts. No custom code was used.

Competing interests

The authors declare no competing interests.