Data on the de novo transcriptome assembly for the migratory bird, the Common quail (Coturnix coturnix)

Abstract

The Common or European quail (Coturnix coturnix) is a Galliform bird of ecological importance for research in the field of animal migration. The Common quail is also a game bird, thus, of great interest for recreational activities and conservation management. Here, we generated a high quality de novo transcriptome for the Common quail for which no reference genome is to date publicly available. The transcriptome was obtained from a population of Common quail originated from captive founders raised under laboratory conditions. Paired-end RNA-Sequencing reads were obtained from extracted total RNA of brain tissue punches (preoptic-hypothalamic region) of 23 quails, which yielded to 5.5–11.2 million reads per individual bird for a total of 236 million reads. After assembly optimization, we used a stringent filtering analysis pipeline to remove redundant and low confidence transcripts. The final transcriptome consisted of 22,293 transcripts of which 21,551 (97%) were provided with annotation data. Our data offers a high quality pipeline for compiling transcriptomes of complex non-genomic species. Our data also provide a robust reference for gene expression studies in this species or other related Galliform species, including the Japanese quail.

Specifications Table

Subject	Biology
Specific subject area	Transcriptomics and neurophysiology
Type of data	Transcriptome assembly, raw sequences
How data were acquired	High-throughput RNA-Sequencing performed on a HiSeq V4 platform at the Vienna Biocenter (Vienna, Austria)
Data format	Raw, Analysed
Parameters for data collection	All experimental birds were raised in captivity on a 16:8 hrs light:dark cycle. Upon young adulthood experimental quails were exposed to a gradual decline in day length (i.e. 30 min/week until the photoperiod reached 12:12 hrs light:dark cycle) to simulate autumn migration followed by a non-migratory stage in the non-breeding grounds. For the present RNA-Sequencing experiment, the assembly de novo was generated by pooling together reads of 23 individual birds (11 birds were sampled during the migratory phase and 12 birds were sampled during the non-migratory phase).
Description of data collection	Total RNA was extracted from brain tissue punches (preoptic-hypothalamic region). A paired-end 125 basepair Next Generation Sequencing library was prepared. Sequencing reads were analysed to obtain a de novo transcriptome assembly of the Common quail (Coturnix coturnix).
Data source location	Birds used in this experiment originated from a captive colony of breeding Common quails maintained at the Konrad Lorenz Institute of Ethology (Vetmeduni Vienna, Austria). All experimental procedures were carried out in laboratory conditions at the Konrad Lorenz Institute of Ethology.
Data accessibility	Raw RNA-Sequencing data and assembly file are available from the Sequence Read Archive (SRA) on NCBI, study accession number PRJNA630138 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA630138/). The associated annotation data are available as Supplementary Material.

Value of the Data

• •In this article, we provide the first published de novo transcriptome for the migratory Galliform, the Common quail (Coturnix coturnix). These data provide an important reference to enhance our understanding of the molecular and physiological pathways associated with animal migration, especially in bird species.
• •The Common quail transcriptome data provided here can be used as a reference for differential expression analysis of RNA-Sequencing data.
• •These data will be fruitful for behavioural neuroendocrinologists, ecologists, and conservation biologists. These data are likely to provide important insights also to veterinarians and biomedical scientists working in the field of nutrition and focusing on studying gene expression pathways linked to rapid changes in body mass and fat stores, including metabolic disorders such as obesity.

1. Data description

Data reported in this article originate from RNA-Sequencing of brain tissue punches (hypothalamus) from 23 Common quails (Coturnix coturnix). Paired-end sequences obtained from each individual bird were subsequently assembled together to obtain the transcriptome assembly of the Common quail.

1.1. Quality check of RNA-Sequencing libraries

RNA-Sequencing libraries generated from hypothalamic samples of Common quail yielded between 5.5 and 11.2 million paired-end reads (on average 9.1 million reads) per individual bird for a total of 233.5 million paired-end reads. Quality filtering removed between 5.1% and 9.1% reads per sample, which produced a subset of high quality RNA-Sequencing paired-end reads containing between 4.3 and 9.5 million paired-end reads per individual (on average 7.7 million reads).

1.2. Trinity parameter optimization of de novo transcriptome assembly

1.2.1. Quality assessment and read representation

We compiled three independent de novo assemblies using different parameters within the Trinity package. As can be seen in Table 1, overall, there were relatively minor differences among the three compiled transcriptomes. The total number of assembled bases ranged between 344,461,191 to 441,854,562 among the three de novo assemblies with a median transcript length between 360 and 380 (Table 1a). A similar number of reads aligned back to the transcriptome in the Trinity run with the default setting and in the Trinity run with target K-mer coverage set to 30 (84.1% and 83.7%, respectively), whereas re-alignment rate was in comparison lower (78.8%) in the Trinity run with K-mer minimum coverage set to 2 (Table 1c). As shown in Table 1c, the largest number of reads that aligned concordantly back to the transcriptome exactly one time was obtained in the Trinity run with minimum K-mer set to 2 (31,659,608 paired reads out of 176,376,164 total paired reads, 18.0%).

Table 1.

Summary Trinity statistics of the three compiled de novo assemblies of the Common quail transcriptome. Data are provided for all transcripts (a) or, for the longest isoform per unigene (b). Summary read mapping results from bowtie2 is provided in (c). T indicates the target K-mer coverage and minimum K-mer indicates the minimum number of K-mers to be included in the initial Inchworm contigs.

	T = 50	T = 30	minimum K-mer = 2
(a) All transcript contigs
Contig N50	1,329	1,355	1,589
Median contig length	370	380	360
Average contig	744.1	759.98	781.53
Total assembled bases	441,854,562	441,167,562	344,461,191
(b) Longest isoform per gene
Contig N50	594	605	637
Median contig length	328	328	304
Average contig	522.25	527.68	524.44
Total assembled bases	225,940,673	217,544,496	155,153,297
(c) Bowtie2
failed alignments	4320,019 (2.45%)	4516,006 (2.56%)	5653,533 (3.21%)
aligned concordantly 1 time	23,815,747 (13.50%)	24,265,641 (13.76%)	31,659,608 (17.95%)
aligned concordantly >1 times	148,240,398 (84.05%)	147,594,517 (83.68%)	139,063,023 (78.84%)

1.2.2. Assessment of assembly completeness

Most of the avian core genes were successfully recovered in all the three Trinity assemblies (Table 2). Specifically, out of a total of 4,915 of the single-copy orthologs between 81% and 84% were recovered completely, and between 7% and 10% were recovered partially. Only between 8% and 11% of the 4,915 single-copy orthologs were classified as missing among the three compiled assemblies. These data indicate very good coverage and high quality of the assemblies of the protein-coding transcriptomes for the Common quail.

Table 2.

Summary of the complete, duplicated, fragmented, and missing orthologs from Benchmarking Universal Single-Copy Orthologs (BUSCO) search against the 4,915 single-copy orthologs for Aves within the three compiled de novo assembly. T indicates the target K-mer coverage and minimum K-mer indicates the minimum number of K-mers to be included in the initial Inchworm contigs.

Type of BUSCO ortholog	T = 50	T = 30	minimum K-mer = 2
Complete	4,130 (84.03%)	3,989 (81.16%)	4,042 (82.24%)
(a) Complete and single-copy	1,394 (28.36%)	1,335 (27.16%)	1,426 (29.01%)
(b) Complete and duplicated	2,736 (55.67%)	2,654 (54.00%)	2,616 (53.22%)
Fragmented	370 (7.53%)	481 (9.79%)	348 (7.08%)
Missing	415 (8.44%)	445 (9.05%)	525 (10.68%)

1.2.3. Removal of redundant transcripts and assembly selection

Depending on the assembly parameters used in the Trinity package, the TransDecoder filtering of redundant transcripts resulted in the reduction of the number of assembled transcripts by 9–11 fold. As can be seen in Table 3, the largest number of transDecoder ORFs was obtained in the assembly with target K-mer coverage set to 30 (54,532 ORFs), intermediate values were obtained in the assembly with default settings (52,272 ORFs), while the lowest value of ORFs was detected in the assembly with minimum K-mer set to 2 (50,345 ORFs).

Table 3.

Summary results from TransDecoder for each type of predicted open reading frame. T indicates the target K-mer coverage and minimum K-mer indicates the minimum number of K-mers to be included in the initial Inchworm contigs.

Type of coding sequence	T = 50	T = 30	minimum K-mer = 2
Complete (both start and stop codons)	33,362 (63.82%)	33,218 (60.91%)	30,495 (60.57%)
5prime partial alignments	10,393 (19.88%)	11,571 (21.22%)	11,328 (22.50%)
3prime partial alignments	5,140 (9.83%)	6,022 (11.04%)	4,925 (9.78%)
Internal alignments	3,377 (6.46%)	3,721 (6.82%)	3,597 (7.14%)
total	52,272	54,532	50,345

Despite the relatively minor differences across the three compiled Trinity assemblies, data gathered by Bowtie2 and TransDecoder suggested that the transcriptome in which target K-mer coverage was set to 30 yielded to a reasonable trade-off between read representation and specificity. Consequently, we selected the latter de novo assembly for all the further RNA-Sequencing analyses in the pipeline.

1.2.4. Clustering of non-redundant transcripts

The non-redundant transcripts in the selected Trinity assembly were clustered together using CD-Hit-EST according to a stringent similarity threshold of 0.95. We thus obtained a total number of 22,293 clusters (41% out of the TransDecoder transcript sets).

1.3. Assembly functional annotation

We were able to annotate the majority of transcripts of the final assembly (97%, 21,551 out of 22,293 transcripts). Out of 21,551 total hits detected, 20,462 hits (95%) belonged to the class Aves. Out of 20,462 total hits belonging to Aves, 11,850 hits (58%) were in common with the chicken (Gallus gallus). The taxonomic distribution for the top five most represented bird species in our annotated final transcriptome is shown in Fig. 1. Several Gene Ontology (GO) terms with a relatively enriched representation of cellular, molecular, and biological processes were also identified (Fig. 2). The distribution of the number of transcripts associated with the 15 most frequently occurring EggNOG identifiers is provided in Table 4.

Fig. 1.

Taxonomic distribution of five most represented species (out of 21,551 hits) detected using FunctionAnnotator.

Fig. 2.

Frequency distribution of Gene Ontology (GO) terms for transcripts of Common quail identified in FunctionAnnotator for each primary category: (a) biological process, (b) molecular function, and (c) cellular component.

Table 4.

Number of non-redundant clustered transcripts assigned to the 15 most occurring EggNOG identifiers.

EggNOG Identifier	Description	N transcripts
COG5048	Zinc finger protein	344
COG0666	Ankyrin Repeat	173
COG4886	Leucine Rich Repeat	170
COG0515	Serine Threonine protein kinase	156
ENOG410XRW9	Receptor	145
COG5069	Microtubule associated monoxygenase, Calponin and LIM domain containing	113
COG5059	Kinesin family member	76
COG5022	Myosin heavy chain	75
ENOG410XQHI	Cadherins are calcium dependent Cell adhesion proteins (By similarity)	75
COG5599	Protein tyrosine phosphatase	66
COG0553	Helicase	64
COG1226	PotAssium voltage-gated channel	63
COG2940	Histone-lysine N-methyltransferase	60
COG1100	GTP-binding Protein	57
COG5021	Ubiquitin protein ligase	57

2. Experimental design, materials, and methods

2.1. Experimental animals

The birds used in this article were obtained from a breeding captive colony of outbred Common quails kept at the Konrad Lorenz Institute of Ethology (Vetmeduni Vienna, Austria). Our stock birds originated from wild founders captured during the spring migration/breeding season on the southern Italian coast near Palermo in 2008, 2009, and 2010 (Istituto Sperimentale Zootecnico per la Sicilia, Palermo, Italia). As previously shown [1], microsatellite and mtDNA screening excluded the presence of admixture with the Japanese quail in our study population. Experimental birds were reared under a 16:8 hrs light:dark cycle since hatching until they were about 7 weeks of age (46–59 days-old). Food and water were provided ad libitum throughout the course of the experiment. The indoor rooms were under constant climate control at 20–24 °C.

2.2. Experimental manipulation of the migratory phenotype

From about 7 weeks of age until about 14 weeks of age, all experimental birds were exposed to a gradual decline in day length (30 min/week) until the photoperiod reached 12:12 hrs light:dark. This photoperiod schedule was then maintained constant until the end of the experiment. This photoperiod manipulation aimed to simulate autumn migration followed by a non-migratory life history stage in the non-breeding grounds. After 8 weeks since the start of the photoperiod decline (average age of the birds was about 15 weeks, 102–114 days-old), half of the birds were randomly chosen and allocated to the migratory sampling group (October 2017), whereas the remaining birds were allocated to the non-migratory group (December 2017). For both groups, sampling procedures were performed at a standardised time of the day. For the RNA-Sequencing experiment presented here, we selected 11 birds sampled during the migratory phase (female: 6, male: 5) and 12 birds sampled during the non-migratory phase (female: 7, male: 5) – all the birds used here were sampled at night time (i.e. 21:00 h, three hours after lights were switched off).

2.3. Brain collection

Birds were sacrificed using an overdose of sodium pentobarbital 200 mg/ml via intraperitoneal administration. Birds were then decapitated, and brains removed after 6 min postmortem on average (4–9 min). The brains were immediately placed on dry ice and stored at −80 °C for further dissections. Punch dissections were performed as previously described in the related Japanese quail [2]. Briefly, the brains were placed ventral side up into a frozen brain holder matrix (Hugo Sachs Elektronik, Germany) with a 1 mm graduated scale. Two blades were positioned ∼ 4 mm from the rostral pole and ∼2 from the cerebellum to obtain a 2mm-tick coronal section. The cutting plane was adjusted to match as closely as possible the plane of the chicken brain atlas ([3] – coronal brain sections interaural 2.08–2.56 mm). Then, one single punch per individual brain was obtained from the basal hypothalamus spanning the third ventricle, and immediately stored at −80 °C until laboratory analysis.

2.4. RNA isolation

Total RNA was extracted using the Rneasy Microarray Tissue Mini Kit (Qiagen, Hilden, Germany), with a DNase digestion step to remove DNA contaminants using RNase-Free DNase Set (Qiagen, Hilden, Germany) following previously validated protocol in the Japanese quail [2]. Both purity and integrity of RNA were assessed respectively using a Nanodrop spectrophotometer (ThermScientific, Waltham, MA, USA) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Hypothalamic RNA concentrations averaged 86.3 0± 5.80 ng/μl, ratio 260/280: 2.01 ± 0.01, Agilent RIN scores averaged 9.03 ± 0.06 (mean ± se for all).

2.5. cDNA library construction and RNA-Sequencing

For each individual hypothalamic sample, 500 ng of high-quality RNA (RIN ≥ 8.3 for all) was used to construct 23 barcoded sequencing libraries using standard Lexogen Sense mRNA-sequencing Library Preparation kit (Lexogen, Vienna, Austria) according to the manufacturer's instructions. The obtained cDNA libraries were then sequenced paired-end in one lane with a sequencing run length of 125 on a HiSeq V4 platform at the Vienna Biocenter (Vienna, Austria).

2.6. Transcriptome data analysis

2.6.1. Standard quality assessment of sequencing reads

The sequencing reads (BAM format) were pre-processed following standard guidelines. After conversion to fastq format, we used the software Trimmomatic v0.38 [4] to remove Illumina adapter sequences (up to two mismatches were allowed when recognizing adaptor sequences) and to filter out low quality reads (average Phred quality score < 30 in a sliding window of 8 base pairs across entire reads and reads below a minimum length of 50). Quality control of processed reads was confirmed and visualized using FastQC v0.11.7 (https://www.bioinformatics.babraham.ac.uk/index.html).

2.6.2. Trinity de novo assembly and parameter optimization

A reference genome for the Common quail is currently not available. Thus, we assembled the transcriptome de novo using the quality filtered paired-end sequencing reads. Assembly de novo strategies can outperform reference genome transcriptome mapping strategies for species with complex genomes [5], [6], [7]. The program Trinity v2.7.0 was used for the de-novo assembly of the pre-processed reads [8]. Briefly, Trinity assembly pipeline consists of three consecutive modules: Inchworm, Chrysalis, and Butterfly. In the first stage, all overlapping K-mers are extracted from the RNA-Seq reads. The Inchworm module then assembles initial contigs by extending sequences with abundant K-mers. In the following stage, Chrysalis clusters overlapping Inchworm contigs, builds de Bruijn graphs for each cluster, and partitions reads among clusters. Finally, Butterfly resolves all probable sequences from each graph component predicting alternatively spliced and paralogous transcripts independently for each cluster. In order to assess software performance with our data we used three independent strategies to compile the transcriptome assembly. Apart from assembly using default parameters (K-mer size set to 25 and target K-mer coverage set to 50, –normalize_max_read_cov), a second assembly was obtained using default parameters but with the target K-mer coverage set to 30. The third assembly was obtained by using all default parameters except for the K-mer minimum coverage (–min_kmer_cov) which was set at 2 in order to exclude all singletons obtained in the initial Inchworm contigs. We compared the three assemblies using standardised metrics following recommendations in the Trinity software (https://github.com/trinityrnaseq/trinityrnaseq/wiki). The de novo assemblies were performed on the Department of Integrative Biology and Evolution server with 4 processors, 80 cores and 378 GB RAM.

2.6.3. Assessment of assembly quality, read content, and assembly completeness

First, we retrieved basic quality metrics statistics included in the Trinity package. These data were obtained for all transcript contigs and also for only the longest contig from each predicted gene. To further assess the quality of the three de novo assemblies we then examined the number of input RNA-Sequencing reads that were represented in each assembled transcriptome. Thus, we mapped cleaned reads back to their corresponding assemblies using Bowtie2 v2.3.1 in local read alignment to maximize the alignment score (–local) and by suppressing records for reads that failed to align (–no unal) [9]. In order to assess transcriptome completeness we also compared our assembled transcripts sets against a set of known highly conserved single-copy orthologs using BUSCO v 3.0.2 (Benchmarking Universal Single-Copy Orthologs). We used the predefined lineage Aves ortholog dataset (i.e. Aves_odb9, Eukaryota) containing 4,915 single-copy orthologs from the OrthoDB database [10]. We consequently calculated the number of complete (length is within two standard deviations of the mean length of the given BUSCO), duplicated (complete BUSCOs represented by more than one transcript), fragmented (partially recovered BUSCOs) and missing (not recovered) transcripts in each of our three Trinity assembly.

2.6.4. Removal of redundant transcripts and hierarchical clustering

We used TransDecoder v5.3.0 ([8]; https://github.com/TransDecoder) to first extract all predicted protein coding sequences and then to select the single-best open reading frame (ORF) for each transcript within our three de novo assemblies (–single_best_orf, default setting). Any transcripts with ORFs with less than 200 bp in length were removed before proceeding with further analyses in the pipeline.

2.6.5. Hierarchical clustering in non-redundant transcripts

Using the selected assembly de novo, the candidate coding regions identified using TransDecoder were further filtered using the software CD-Hit-Est v4.6.8 [11]. CD-Hit-Est clustered highly similar nucleotide sequences that meet a sequence identity threshold of 0.95 (default = 0.90).

2.6.7. Functional annotation

In order to obtain a comprehensive annotation of our final sets of transcripts we used FunctionAnnotator, which is especially well-suited for non-model organism annotation [12]. The uploaded final transcriptome of Common quail in FunctionAnnotator was used for homology searches against the NCBI-NR (Non-redundant) protein sequences database with a cut-off e-value of 1e⁻³. Further functional annotations to explore distribution and frequencies of KEGG (Kyoto Encyclopedia of Genes and Genomes; [13]) and eggNOG [14] was performed using Trinotate v3.2.0 [15].

Ethical statement

All animal procedures were performed in compliance with the Austria legislation with approval of the Ethics Committee of the University of Veterinary Medicine Vienna, and the Federal Ministry of Science, Research and Economy (BMWFW-68.205/0037-WF/V/3b/2017).

Funding information

The work was funded by a Marie Sklodowska-Curie Individual Fellowshipto to VM (#704582) and by intramural funds to LF. VM was additionally supported by a FWF Der Wissenschaftsfonds Lise Meitner Fellowship(#M2520-B29).

Author contribution

VM designed the study, performed the animal work, contributed in supervision of laboratory work, performed data analysis, and wrote the manuscript. LF gave input on study design and provided comments on earlier drafts of the manuscript. GP established the colony of captive Common quails from which the birds used in this experiment originated and contributed to the planning of the experiments. SS gave input on study design, supervised laboratory and data analyses, and provided comments on earlier drafts of the manuscript.

Declaration of Competing Interest

We declare no competing interests.