Keywords: bivalve, genome sequencing, 2b-RAD genotyping, linkage mapping, quantitative trait locus mapping

2.1. Genome survey sequencing and gene annotation

A two-year-oldC. farreri individual was chosen for genome survey sequencing. Genomic DNA was extracted from its adductor muscle using the standard phenol/chloroform extraction method.²³ Three paired-end DNA libraries with insert sizes of 175, 500 and 800 bp were constructed by following the Illumina's standard DNA library preparation protocol and were then sequenced using the Illumina HiSeq2000 platform.

Raw reads were first filtered to remove low-quality reads resulting from base-calling duplications or adapter contamination. Clean reads were assembled using the SOAPdenovo software, which applies thede Bruijn graph structure to construct contigs. Subsequently, all reads were realigned to the contigs and the paired-end information was used to join unique contigs into scaffolds. Finally, intra-scaffold gaps were filled using paired-end extracted reads that had one read uniquely aligned on a contig and another read located in a gap region. Genomic sequences were archived in the Sequence Read Archive (SRA) database (accession no. SRP018107).

Our group has recently reported the most comprehensive transcriptome resource forC. farreri by 454 sequencing of cDNA libraries generated from diverse developmental stages and adult tissues,⁷ which provide an excellent basis for gene annotation of the obtained scaffolds. Genomic scaffolds were first annotated by aligning the 20 056 transcriptomic isogroup sequences (i.e. the longest isotig in each isogroup) onto these scaffolds using the GMAP program.²⁴ To increase the annotation rate, the unannotated scaffolds were compared against the Swiss-Prot and non-redundant (NR) protein databases using BlastX, with anE-value threshold of 1E−4. For scaffolds without significant matches, tBlastX search with anE-value threshold of 1E−4 was conducted against the NR nucleotide database for further annotation. To increase computational speed, all Blast searches were limited to the top 10 significant hits for each query.

2.2. Mapping families

Twenty full-sib families were established in May 2009 by single-pair mating of ‘Penglai-Red’ scallops, an eliteC. farreri strain developed by our group with features of red shell, fast growth and disease resistance. Growth-related traits including shell length, shell width, shell height and body weight were measured for all families at the age of 15 months. One of these families exhibiting high within-family variation of growth traits was chosen for linkage and QTL analysis.

2.3. 2b-RAD sequencing

2b-RAD libraries were prepared for two parents and 96 progenies by following the protocol developed by Wanget al.² For the parents, standard BsaXI libraries were constructed, whereas for the progenies, reduced representation (RR) libraries were constructed using adaptors with 5′-NNT-3′ overhangs to target a subset of all BsaXI fragments in theC. farreri genome. A unique barcode was incorporated into each library during library preparation, and then all libraries were pooled for single-end sequencing (1 × 50 bp) using an Illumina GA-II sequencer. All the 2b-RAD sequences were archived in the SRA database (accession no. SRA065207).

2.4. Sequence data preprocessing andde novo genotyping

Raw reads were first trimmed to remove adaptor sequences. The terminal 3-bp positions were excluded from each read to eliminate artefacts that might have arisen at ligation sites. Reads with no restriction sites or containing ambiguous base calls (N), long homopolymer regions (>10 bp), excessive numbers of low-quality positions (>5 positions with quality of <10) or mitochondrial origins were removed. The remaining trimmed, high-quality reads formed the basis for subsequent analysis.

De novo 2b-RAD genotyping was performed using the RADtyping program v1.0 (http://www2.ouc.edu.cn/mollusk/detailen.asp?Id=727) under default parameters. The RADtyping program has recently been developed by our group, and it is an integrated pipeline that enables both accuratede novo codominant and dominant genotyping in mapping populations.

2.5. Linkage map construction

Segregating markers that could be genotyped in at least 80% of the individuals were considered as high-quality markers and were retained for further analysis. For each segregating marker, goodness of fit of the observed with the expected Mendelian ratio was assessed using theχ² test, and those markers conforming to the expected Mendelian ratios (P ≥ 0.05) were included in map construction. Linkage analysis was performed using the JoinMap 4.0 software.²⁵ Sex-specific maps were first constructed for each parent. Maternal and paternal datasets were created using the function ‘Create Maternal and Paternal Population Nodes’ in the JoinMap program, which was also used to partition 1:2:1-type data into 1:1 female-type and 1:1 male-type data. A logarithm of odds (LOD) threshold of at least 6.0 was used to assign markers into linkage groups. The regression mapping algorithm was selected for map construction. The Kosambi mapping function was used to convert the recombination frequencies into map distances (centiMorgans). A consensus map was established by integrating the sex-specific maps through shared markers using the MergeMap software.²⁶

2.6. QTL mapping and association analyses of growth traits and sex

The distributions of growth traits were first assessed using the univariate procedure of MATLAB software to check whether they followed normal distributions. Pearson correlations among four growth traits were calculated using pairwise comparison.

The regression method was used for QTL mapping of both growth traits and sex. Here, sex was used as a binary trait, supposing the haplotypes of the male and female parents are H_pp, H_pm, H_mp and H_mm, thata_f,−a_f,a_m and−a_m are their respective effects on the trait of interest, andP_pp,P_pm,P_mp andP_mm are the respective individual inheritance probabilities at putative positions in the corresponding parental haplotypes. Following the above assumption,P_pp +P_pm =P_mp +P_mm = 1. Therefore, the hypothesis to detect each QTL is:

(1)

and its null hypothesis is:

(2)

wherey is the observed value of the trait of interest;μ, the population mean;s, the fixed effect of sex ande, the random error.

For each chromosome, midpoints of all marker brackets along the chromosome were considered to be putative QTL positions, and the inheritance probabilities for individuals at each position were calculated using the IBD program²⁷ and the corresponding LOD value was determined by:

(3)

where the LRT statistic was calculated through the comparison of Equations (5) and (6). Chromosome-wide and genome-wide critical threshold values for QTL detection were approximated using Piepho's method.²⁸ If a QTL was detected, the confidence interval was calculated using Li's method.²⁹

Association analysis was performed as a complement approach to QTL mapping, which was implemented using an R package called ‘GWAF’³⁰ that tests genetic association between genotypes and traits by fitting a linear model accounting for within full-sib pedigree correlation. The additive model in GWAF was selected by use of the kinship coefficient matrix in the analysis to avoid false positives owing to unexpected familial correlation. The functions ‘lme.batch’ and ‘gee.lgst.batch’ were used to perform a global test (i.e. Waldχ² test) of genotypic effects with growth traits and sex, respectively. Bonferroni correction was carried out to account for multiple comparisons.

2.7. Integration of the linkage map, genomic scaffolds and a physical map

To associate genomic scaffolds with the linkage map, all markers in the linkage map were mapped onto the scaffolds using the SOAP2 software³¹ (parameters –M 4, -v 2). Marker-associated scaffolds were further used as linkers to integrate the linkage map with a BAC-based physical map previously constructed forC. farreri.¹⁸ These scaffold sequences were compared against available BAC-end sequences (BESs) using BlastN with anE-value cut-off of 1E−10. A reciprocal comparison was performed to ensure one-to-one matches between the scaffolds and BAC contigs.

3.1. Genome survey sequencing ofC. farreri

The genome size ofC. farreri is relatively large and was previously estimated as ∼1.2 Gb.²⁰ Genome survey sequencing was performed for a singleC. farreri individual by sequencing three paired-end DNA libraries with insert sizes of 175, 500 and 800 bp based on the Illumina HiSeq2000 platform. In total, 62.4 Gb of sequencing data, equivalent to ∼52× genome coverage, were produced (Table 1).K-mer analysis revealed remarkably high heterozygosity (∼1.4%) in theC. farreri genome (Fig. 1). Owing to the high genome heterozygosity, assembly of these data produced a large number of small scaffolds (N50 = 1.5 kb), with the longest one only reached to 46 kb (Table 2). All the scaffold sequences are accessible athttp://ipl.ouc.edu.cn/fuxiaoteng/cf_SRA_data. Of 20 056 isogroups in theC. farreri transcriptome,⁷ 18 316 (91.3%) could match with 16 489 genomic scaffolds, suggesting that the majority of genes were present in our reference genome dataset. Of the mapped transcriptomic isogroups, 88.4% had been functionally annotated through the homology-based comparison against public databases. Summing up all scaffolds resulted in a total length of ∼870 Mb, corresponding to ∼70% of genome coverage; though not complete, it constitutes the most informative reference genome currently available forC. farreri.

3.2. Optimization of a 2b-RAD sequencing plan for linkage mapping

Because of the large genome size (∼1.2 Gb) ofC. farreri, it is necessary to devise a cost-effective sequencing plan that can balance the total sequencing cost and genotyping accuracy. In total, 210 570 BsaXI sites were identified from the reference genome dataset. Based on the genome coverage (∼70%) of this dataset, the total number of BsaXI sites in theC. farreri genome was estimated as ∼0.3 million. Sequencing all of the BsaXI sites to 20× for a large mapping population (e.g. hundreds of individuals) would require a large sequencing investment. A notable feature of the 2b-RAD method is its ability to fine tune marker density as needed by constructing RR libraries using less-degenerate adaptors.² For example, an RR library constructed using adaptors with 5′-NNT-3′ overhangs only targets ∼1/10th of all the BsaXI sites in theC. farreri genome, and sequencing of such a library to 20× would require only 0.6 million reads per individual (in contrast to 6 million reads per individual for a standard BsaXI library), thus dramatically reducing the total sequencing cost.

3.3. 2b-RAD sequencing of aC. farreri mapping population

Based on the calculations above, RR libraries were constructed for aC. farreri mapping population, sequencing of which produced 0.6–1.2 million high-quality reads per progeny. For the parents, standard libraries were constructed and sequenced to a sufficient depth (70–80×). As shown in Fig. 2, target sites of the RR libraries could be reproducibly detected among different progenies and their average sequencing coverage was ∼6-fold higher than that of standard libraries. The sequencing depth for progenies ranged from 13.4 to 23.8 with an average of 16.75. Clustering parental reads resulted in 181 625 representative reference sites consisting of 138 141 parent-shared sites and 43 484 parent-specific sites. After filtering low-quality sites, 117 113 parent-shared and 35 799 parent-specific sites remained. These reference sites contained 92% of the unique BsaXI sites inferred from the scaffolds, suggesting that unique sites were well represented in the high-quality reference sites.

3.4. High-resolution linkage mapping

In total, 7458 polymorphic markers were identified, of which 6842 were heterozygous in at least one parent, including 2196 co-dominant and 4646 dominant markers. Dominant markers showing 3 : 1 segregation pattern in progenies were not included in subsequent analysis due to their inability to construct sex-specific maps. Totally, 4881 markers that conformed to the expected Mendelian ratios (P ≥ 0.05) were included in the linkage analysis. These markers were partitioned into maternal (1 : 1 female type; 2473 markers) and paternal (1 : 1 male type; 2175 markers) datasets for constructing sex-specific linkage maps. At the LOD threshold of 6.0, the markers were grouped into 19 linkage groups, corresponding to the haploid chromosome number ofC. farreri.³² The female map contained 2025 markers and spanned 1175.4 cM with an average marker interval of 0.59 cM, whereas the male map contained 1861 markers and spanned 1154.9 cM with an average marker interval of 0.62 cM. Significant differences in recombination rates between the female and male maps were observed for linkage groups LG5, LG8, LG10 and LG15 (Table 3). LG5 and LG10 showed higher recombination rates in females than in males, whereas the contrary was found for LG8 and LG15.

The sex-specific maps were further integrated using anchor markers that were heterozygous in both parents. The consensus map contained 3806 markers (Fig. 3, Table 3) and spanned 1543.4 cM with an average marker interval of 0.41 cM. The length of each linkage group ranged from 54.9 to 99.6 cM, and the marker density varied from 123 to 274 across the linkage groups. Mapping all the markers to the annotated genomic scaffolds revealed 623 markers (18–53 per linkage group; Table 3) that resided in or close to genes. These markers represent a valuable resource for future comparative genome analysis.

Genome length was estimated to be 1544.2 cM (G_e1) and 1559.5 cM (G_e2) using two different methods.^33,34 The average of the two estimates (1551.9 cM) was used as the expected genome length. Genome coverage of the consensus map was nearly complete and reached 99.5%. Given the estimated genome size of ∼1.2 Gb forC. farreri,²⁰ the average recombination rate across all the linkage groups was ∼1.3 cM/Mb.

3.5. QTL mapping and association analyses of growth traits and sex

Pairwise comparisons among five growth traits using Pearson's correlation revealed that all of them were highly correlated with correlation coefficients ranging from 0.77 to 0.96 (Table 4). The strongest correlation was observed between shell length and shell height (r = 0.96,P = 1E−10). A genome scan for these growth traits showed that they exhibited quite similar LOD profiles (Fig. 4), indicating that these traits were possibly regulated by the same set of genes. Therefore, a principle component analysis was conducted, which revealed that the first principal component (PC1) explained ∼90% of the total phenotypic variation (Fig. 5). To increase the QTL detection power, PC1 was utilized as a composite trait for QTL mapping. Two significant QTLs were detected on LG1 and LG3 (Fig. 6a and Table 5) and the one on LG3 that passed the genome-wide significance threshold showed a stronger QTL signal. The two QTLs explained 11.4 and 16.9% of the total phenotypic variation, respectively. Given the high marker density in the linkage map, association analysis was also performed to serve as a complementary approach to evaluate the QTL mapping results. Association analysis revealed a similar distribution pattern across all linkage groups as in QTL mapping analysis (Fig. 6a) and supported QTL mapping results in general. Within the confidence interval of the QTL detected on LG3, one marker named f68558 was found to be located in the intron region of a transcription factor gene,PROP1 (Homeobox protein prophet of Pit-1) (Fig. 7 andSupplementary Table S1). It has been shown thatPROP1 can regulate the production of growth hormone,³⁵ a peptide that simulates growth, cell reproduction and regeneration in animals. Identification ofPROP1 within the QTL region suggests that this gene may represent an important candidate QTL gene that is worthy of further investigation.

While for sex, a highly significant QTL was finely mapped on LG11 with the confidence interval of 0.37 cM (34.61–34.98 cM; Fig. 6b). Moreover, no global difference (female : male = 1.01) in the recombination rate was observed for this linkage group between the two sex-specific maps (Table 3). Association analysis revealed 27 sex-related markers with high statistical significance (P < 1E−6;Supplementary Table S2) and all of them fell into the narrow QTL region identified by the QTL mapping analysis. Interestingly, one sex-related marker named f93422 was located in the coding region of a transcription factor gene,ZNFX1 (NFX1-type zinc finger-containing 1), which has been shown to be tightly linked withAMHR2 (anti-Mullerian hormone receptor type II), the sex-determination gene in the tiger pufferfish,Takifugu rubripes.³⁶

3.6. Integration of the linkage map, genomic scaffolds and a physical map

A high-resolution linkage map can serve as an important platform for integrating multiple genomic resources. One such application is to improve genome assembly by orienting genomic scaffolds. ForC. farreri, a majority of larger genomic scaffolds contained at least one BsaXI site (77% for scaffolds of > 5kb; 92% for scaffolds of >10 kb) and therefore, it could be potentially oriented using a high-density linkage map. Of the 3806 markers in the consensus linkage map, 2174 could be unambiguously anchored to scaffolds. These scaffolds were not only useful for directing a higher-level genome assembly, but also could serve as ‘long’ surrogates for 2b-RAD tags to enhance their utility in unifying genomic resources. To demonstrate this potential, we attempted to integrate the linkage map with a BAC-based physical map previously constructed forC. farreri.¹⁸ The physical map contained 2514 BAC contigs that had BESs; however, a large fraction of them had only a few BESs (e.g. 73% with less than three sequenced BAC clones), thus limiting the integration efficiency. Even so, 681 BAC contigs could still be linked with the linkage map through sequence comparison between BESs and marker-associated scaffolds (Supplementary dataset). An example showing the integration of LG1 and BAC contigs is present in Fig. 8.

4.1. Genome sequencing in bivalves

In spite of species abundance and diverse geographical distribution, little effort has been devoted to decoding bivalve genomes. To date, whole-genome sequencing has been performed for only two bivalve species, i.e. Pearl oyster (Pinctada fucata)³⁷ and Pacific oyster (Crassostrea gigas),³⁸ providing the first glimpse into bivalve genome architecture. Apparently, genome sequencing of a broader range of bivalves would enable a better understanding of their phylogeny, speciation and diversification. Our study represents the first step towards fully decoding a scallop genome. Genome analysis revealed thatC. farreri possesses a highly heterozygous genome. This finding is not unusual and has been reported in other bivalve species.^37,38 For example, it has been shown that, for the Pacific oyster, average density of SNPs is one SNP per 60 bp in coding regions and one per 40 bp in non-coding regions.³⁹ High genome heterozygosity poses the challenge of efficient genome assembly. In such instance, a high-resolution linkage map would be much valuable in that they can orientate the small scaffolds to improve genome assembly.

4.2. Devising an optimal 2b-RAD sequencing plan for linkage mapping

Large-scale linkage mapping studies generally call for a cost-effective sequencing plan that reasonably balances the sequencing cost and genotyping accuracy. Through simulation analysis, a sequencing depth of 20× for both parents and progenies can provide sufficiently high genotyping accuracy (>96%; unpublished data). Determining the amount of sequencing necessary for a desired sequencing depth would require the knowledge of a total number of restriction sites in a given genome. Without a reference genome, the total number of restriction sites can be simply predicted based on genome size and GC content by assuming sites that are Poisson distributed. However, it has been shown that such estimation departed considerably from the actual numbers in some instances.¹ Our study demonstrates that genome survey sequencing can facilitate devising an optimal 2b-RAD sequencing plan by providing sufficient genome information to determine the minimal amount of sequencing required for a given sequencing depth. Meanwhile, the obtained genome information can also be used to enhancede novo 2b-RAD analysis in many other aspects, such as evaluation of the efficiency and reliability ofde novo reference site reconstruction, identification of genes surrounding 2b-RAD tags and association of 2b-RAD tags with public genomic resources.

For species with very large genomes, sequencing all BsaXI sites at a depth of 20× for all individuals remains a substantial investment. For example, sequencing 200 human individuals with the genome size of ∼3 Gb would require 5.2 billion reads, which are approximately equivalent to the number of reads produced from >4 full sequencing runs using the HiSeq2000 platform. One advantage of 2b-RAD methodology is the option to adjust marker density to the desired level using selective adaptors. Sequencing cost can be further reduced by sequencing a subset of BsaXI sites deriving from RR libraries. We have demonstrated that sequencing aC. farreri RR library only required about one-sixth of the sequencing cost for a standard library at a given sequencing coverage, thus reducing the total sequencing cost dramatically. Therefore, sequencing RR libraries represents an effective option for large-scale linkage mapping studies dealing with species with large genomes.

4.3. Linkage map construction and QTL analysis

Thus far, linkage maps have been constructed for many bivalve species includingC. farreri.^14–17 However, these maps were usually built using hundreds of markers and the resolution was generally low (mostly 10–20 cM), limiting their use in fine-scale QTL mapping and many other applications. In this study, the linkage map constructed forC. farreri contained 3806 markers, a marker density that has, to our knowledge, never been reached for any other bivalve species. This linkage map covered nearly the whole genome (99.5%) with a resolution of 0.41 cM. In addition, ∼10% of the markers in the linkage map residing in or close to genes as revealed by mapping them to the annotated genomic scaffolds. Our map, therefore, provides sufficient resolution for fine-scale QTL mapping and facilitates the discovery of quantitative trait genes.

Growth traits are of particular interest to scallop researchers due to their high commercial significance in scallop aquaculture. QTL mapping represents an efficient approach to identify genetic loci underlying these traits for marker-assisted selection in genetic breeding. Two growth-related QTLs were detected on the linkage groups LG1 and LG3, which are also supported by the association analysis, a complementary approach to evaluate the QTL mapping results. The markers located at the confidence intervals of these QTLs constitute a valuable marker set for further evaluation of their utility in marker-assisted selection. In particular, the homeotic genePROP1, which was identified from the QTL region on LG3, drew our great interest. This gene encodes a paired-like homeodomain protein that is necessary forPIT1 (growth hormone factor 1) expression. In mammals, mutations in this gene can result in impaired production of growth hormone and other pituitary hormones.³⁵ Polymorphisms in this gene have been associated with production traits in livestock animals, such as cattle,⁴⁰ goats^41,42 and sheep.⁴³ The marker f68558 is probably not the causal locus that is responsible for growth trait variation inC. farreri. Instead, we think it is more likely that f68558 is in tight linkage disequilibrium with the causal mutation that is possibly located in the regulatory region (e.g. promoter) ofPROP1. The expression level ofPROP1 has been checked in major adult tissues (e.g. adductor muscle, digestive gland, gill, mantle, male and female gonads and kidney) ofC. farreri. It turned out thatPROP1 was not expressed in neither of these tissues (data not shown), suggesting that scallopPROP1 is probably expressed and functioning during early embryonic development just like its homologues in vertebrates. In vertebrates,PROP1 is only temporarily expressed during early pituitary development.⁴⁴ However, determining the temporal and spatial expression ofPROP1 in bivalves is not straightforward, since the counterpart of pituitary has not been established in bivalves. Further work would be needed to overcome these barriers to evaluate the role(s) ofPROP1 in scallop growth regulation.

Unlike many other animals, scallops do not possess recognizable heteromorphic sex chromosomes,⁴⁵ and little is currently known about their sex-determination mechanism. The use of molecular markers in mapping experiments to identify regions involved in sex determination represents an efficient approach to study sex-determination systems in species without heteromorphic sex chromosomes. In this study, both QTL mapping and association analysis revealed a highly significant sex-related QTL on LG11, suggesting the presence of a narrow sex-determination region (34.61–34.98 cM) inC. farreri. No global differences in recombination rates were detected for LG11 between the two sexes, which may account for the indistinguishable morphologies of sex chromosomes inC. farreri. Interestingly, we found that one sex-related marker (i.e. f93422) was situated at the transcription factor gene,ZNFX1, which has been shown to be tightly linked with the sex-determination geneAMHR2 in the tiger pufferfish.³⁶ Unfortunately, we do not have direct evidence to support the physical linkage betweenZNFX1 andAMHR2 based on the availableC. farreri genome information. Furthermore, searching another two molluscan genomes, i.e. the Pacific oysterC. gigas (http://gigadb.org/pacific_oyster) and the owl limpetLottia gigantea (http://metazoa.ensembl.org/Lottia_gigantea), did not support the two genes' linkage either, though this does not necessarily exclude the possibility of their physical linkage because these genome assemblies are still largely fragmented (scaffold N50 is 401 kb forC. gigas³⁸ and 1.87 Mb forL. gigantean⁴⁶). Our group has initiated the whole-genome sequencing project forC. farreri; therefore, this issue will be settled in the near future.

4.4. Integration of the linkage map with multiple genomic resources

Recent studies have shown that RAD-based linkage maps can greatly improve large-scale genome assembly.^47,48 However, RAD tags are usually short in length (35–100 bp), thus limiting their direct use for integration with other important genomic resources generated from RNA-Seq, ChIP-Seq, exome sequencing or BAC–fosmid-end sequencing. Here, we propose that RAD tags can be effectively extended using the scaffolds obtained from genome survey sequencing. ForC. farreri, although a majority of the scaffolds were not very long (scaffold N50 = 1.5 kb), they could already serve as ‘long’ surrogates for 2b-RAD tags to enhance their utility. We have demonstrated that this approach could facilitate the integration of the scallop linkage map with the BAC-based physical map, even though a large fraction of BAC contigs contained only a few BESs (e.g. 73% with less than three sequenced BAC clones). Besides, such approach can also facilitate the identification of genes surrounding a given tag, which can be crucial for the identification of quantitative trait genes in QTL mapping and association studies. Therefore, it can be envisioned that this scallop linkage map would serve as an important platform for unifying genomic resources that have been accumulated forC. farreri.

4.5. Conclusions

We generated a preliminary reference genome for a bivalve mollusc, expanding the genome resources currently available for bivalves—a group of animals that remain largely unexplored in terms of genome sequencing. A high-resolution linkage map was constructed forC. farreri with a marker density that has, to our knowledge, never been achieved in any other molluscs. Two growth-related QTLs and one putative sex-determination region were detected, and candidate genes identified from these QTL regions represent important targets for further evaluation. Integration of the genome map, linkage map and physical map exemplifies how to build up a comprehensive genomic framework in a non-model organism.

• 1.Davey J.W., Blaxter M.L. RADSeq: next-generation population genetics. Brief Funct. Genomics. 2011;9:416–23. doi: 10.1093/bfgp/elq031. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 2.Wang S., Meyer E., McKay J., Matz M.V. 2b-RAD: a simple and flexible method for genome-wide genotyping. Nat. Methods. 2012;9:808–10. doi: 10.1038/nmeth.2023. [DOI] [PubMed] [Google Scholar]
• 3.Peterson B.K., Weber J.N., Kay E.H., Fisher H.S., Hoekstra H.E. Double digest RADseq: an inexpensive method for De Novo SNP discovery and genotyping in model and non-model species. PLoS One. 2012;7:e37135. doi: 10.1371/journal.pone.0037135. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 4.Appeltans W., Ahyong S.T., Anderson G., et al. The magnitude of global marine species diversity. Curr. Biol. 2012;22:2189–202. doi: 10.1016/j.cub.2012.09.036. [DOI] [PubMed] [Google Scholar]
• 5.FAO. The State of World Fisheries and Aquaculture 2010. Rome: FAO Fisheries and Aquaculture Department, FAO; 2010. [Google Scholar]
• 6.Hou R., Bao Z., Wang S., et al. Transcriptome sequencing and de novo analysis for Yesso scallop (Patinopecten yessoensis) using 454 GS FLX. PLoS One. 2011;6:e21560. doi: 10.1371/journal.pone.0021560. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 7.Wang S., Hou R., Bao Z., et al. Transcriptome sequencing of Zhikong scallop (Chlamys farreri) and comparative transcriptomic analysis with Yesso scallop (Patinopecten yessoensis) PLoS One. 2013;8:e63927. doi: 10.1371/journal.pone.0063927. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 8.Clark M.S., Thorne M.A., Vieira F.A., Cardoso J.C., Power D.M., Peck L.S. Insights into shell deposition in the Antarctic bivalve Laternula elliptica: gene discovery in the mantle transcriptome using 454 pyrosequencing. BMC Genomics. 2010;11:362. doi: 10.1186/1471-2164-11-362. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Meyer E., Manahan D.T. Gene expression profiling of genetically determined growth variation in bivalve larvae (Crassostrea gigas) J. Exp. Biol. 2010;213:749–58. doi: 10.1242/jeb.037242. [DOI] [PubMed] [Google Scholar]
• 10.Ghiselli F., Milani L., Chang P.L., et al. De novo assembly of the Manila clam Ruditapes philippinarum transcriptome provides new insights into expression bias, mitochondrial doubly uniparental inheritance and sex determination. Mol. Biol. Evol. 2011;29:771–86. doi: 10.1093/molbev/msr248. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 11.Guo H., Bao Z., Li J., et al. Molecular characterization of TGF-β type I receptor gene (Tgfbr1) in Chlamys farreri, and the association of allelic variants with growth traits. PLoS One. 2012;7:e51005. doi: 10.1371/journal.pone.0051005. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 12.Wang C., You Y., Wang H., Liu B. Genetic diversity of the sulfotransferase-like gene and one nonsynonymous SNP associated with growth traits of clam, Meretrix meretrix. Mol. Biol. Rep. 2012;39:1323–31. doi: 10.1007/s11033-011-0865-4. [DOI] [PubMed] [Google Scholar]
• 13.Breton S., Stewart D.T., Shepardson S., et al. Novel protein genes in animal mtDNA: a new sex determination system in freshwater mussels (Bivalvia: Unionoida)? Mol. Biol. Evol. 2011;28:1645–59. doi: 10.1093/molbev/msq345. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Wang S., Bao Z., Pan J., et al. AFLP linkage map of an intraspecific cross in Chlamys farreri. J. Shellfish Res. 2004;23:491–9. [Google Scholar]
• 15.Li L., Xiang J.H., Liu X., Zhang Y., Dong B., Zhang X.J. Construction of AFLP-based genetic linkage map for Zhikong scallop, Chlamys farreri Jones et Preston and mapping of sex-linked markers. Aquaculture. 2005;245:63–73. [Google Scholar]
• 16.Wang L., Song L., Chang Y., Xu W., Ni D., Guo X. A preliminary genetic map of Zhikong scallop (Chlamys farreri Jones et Preston 1904) Aquaculture Res. 2005;36:643–53. [Google Scholar]
• 17.Zhan A., Hu J., Hu X., et al. Construction of microsatellite-based linkage maps and identification of size-related quantitative trait loci for Zhikong scallop (Chlamys farreri) Anim. Genet. 2009;40:821–31. doi: 10.1111/j.1365-2052.2009.01920.x. [DOI] [PubMed] [Google Scholar]
• 18.Zhang X., Zhao C., Huang C., et al. A BAC-based physical map of Zhikong scallop (Chlamys farreri Jones et Preston) PLoS One. 2011;6:e27612. doi: 10.1371/journal.pone.0027612. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 19.Zhang L., Bao Z., Wang S., Hu X., Hu J. FISH mapping and identification of Zhikong scallop (Chlamys farreri) chromosomes. Mar. Biotechnol. 2008;10:151–7. doi: 10.1007/s10126-007-9045-x. [DOI] [PubMed] [Google Scholar]
• 20.Zhang L., Bao Z., Cheng J., et al. Fosmid library construction and initial analysis of end sequences in Zhikong scallop (Chlamys farreri) Mar. Biotechnol. 2007;9:606–12. doi: 10.1007/s10126-007-9014-4. [DOI] [PubMed] [Google Scholar]
• 21.Zhang Y., Zhang X., Scheuring C.F., et al. Construction and characterization of two bacterial artificial chromosome libraries of Zhikong scallop, Chlamys farreri Jones et Preston, and identification of BAC clones containing the genes involved in its innate immune system. Mar. Biotechnol. 2008;10:358–65. doi: 10.1007/s10126-007-9071-8. [DOI] [PubMed] [Google Scholar]
• 22.Wang L., Song L., Zhao J., et al. Expressed sequence tags from the Zhikong scallop (Chlamys farreri): discovery and annotation of host-defense genes. Fish Shellfish Immunol. 2009;26:744–50. doi: 10.1016/j.fsi.2009.03.002. [DOI] [PubMed] [Google Scholar]
• 23.Maniatis T., Fritsch E.F., Sambrook J. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1989. [Google Scholar]
• 24.Wu T.D., Watanabe C.K. GMAP: a genomic mapping and alignment program for mRNA and EST sequences. Bioinformatics. 2005;14:1859–75. doi: 10.1093/bioinformatics/bti310. [DOI] [PubMed] [Google Scholar]
• 25.Stam P. Construction of integrated genetic linkage maps by means of a new computer package: Joinmap. Plant J. 1993;3:739–44. [Google Scholar]
• 26.Wu Y., Close T.J., Lonardi S. Accurate construction of consensus genetic maps via integer linear programming. IEEE/ACM Trans. Comput. Biol. Bioinform. 2011;8:381–94. doi: 10.1109/TCBB.2010.35. [DOI] [PubMed] [Google Scholar]
• 27.Sorensen P., Thomsen H. Using DMU for QTL mapping. Foulum, Denmark: Aarhus University; 2003. [Google Scholar]
• 28.Piepho H.P. A quick method for computing approximate thresholds for quantitative trait loci detection. Genetics. 2001;157:425–32. doi: 10.1093/genetics/157.1.425. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 29.Li H. A quick method to calculate QTL confidence interval. J. Genet. 2011;90:355–60. [PubMed] [Google Scholar]
• 30.Chen M.H., Yang Q. GWAF: an R package for genome-wide association analyses with family data. Bioinformatics. 2010;26:580–1. doi: 10.1093/bioinformatics/btp710. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 31.Li R., Yu C., Li Y., et al. SOAP2: an improved ultrafast tool for short read alignment. Bioinformatics. 2009;25:1966–7. doi: 10.1093/bioinformatics/btp336. [DOI] [PubMed] [Google Scholar]
• 32.Wang M.L., Zheng J.S., Yu H. The karyotyping of C. farreri. J. Ocean Univ. Qingdao. 1990;20:81–5. [Google Scholar]
• 33.Chakravarti A., Lasher L.K., Reefer J.E. A maximum likelihood method for estimating genome length using genetic linkage data. Genetics. 1991;128:175–82. doi: 10.1093/genetics/128.1.175. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 34.Fishman L., Kelly A.J., Morgan E., Willis J.H. A genetic map in the Mimulus guttatus species complex reveals transmission ratio distortion due to heterospecific interactions. Genetics. 2001;159:1701–16. doi: 10.1093/genetics/159.4.1701. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 35.Wu W., Cogan J.D., Pfaffle R.W., et al. Mutations in PROP1 cause familial combined pituitary hormone deficiency. Nat. Genet. 1998;18:147–9. doi: 10.1038/ng0298-147. [DOI] [PubMed] [Google Scholar]
• 36.Kamiya T., Kai W., Tasumi S., et al. A trans-species missense SNP in Amhr2 is associated with sex determination in the tiger pufferfish, Takifugu rubripes (fugu) PLoS Genet. 2012;8:e1002798. doi: 10.1371/journal.pgen.1002798. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 37.Takeuchi T., Kawashima T., Koyanagi R., et al. Draft genome of the pearl oyster Pinctada fucata: a platform for understanding bivalve biology. DNA Res. 2012;19:117–30. doi: 10.1093/dnares/dss005. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Zhang G., Fang X., Guo X., et al. The oyster genome reveals stress adaptation and complexity of shell formation. Nature. 2012;490:49–54. doi: 10.1038/nature11413. [DOI] [PubMed] [Google Scholar]
• 39.Sauvage C., Bierne N., Lapègue S., Boudry P. Single nucleotide polymorphisms and their relationship to codon usage bias in the Pacific oyster Crassostrea gigas. Gene. 2007;406:13–22. doi: 10.1016/j.gene.2007.05.011. [DOI] [PubMed] [Google Scholar]
• 40.Thomas M.G., Enns R.M., Shirley K.L., Garcia M.D., Garrett A.J., Silver G.A. Associations of DNA polymorphisms in growth hormone and its transcriptional regulators with growth and carcass traits in two populations of Brangus bulls. Genet. Mol. Res. 2007;6:222–37. [PubMed] [Google Scholar]
• 41.Lan X., Pan C., Zhang L., Zhao M., Zhang C., Lei C., Chen H. A novel missense (A79V) mutation of goat PROP1 gene and its association with production traits. Mol. Biol. Rep. 2009;36:2069–73. doi: 10.1007/s11033-008-9418-x. [DOI] [PubMed] [Google Scholar]
• 42.Xu T.S., Liu J.B., Yao D.W., Cai H.F., Chen H., Zhou H.L., et al. The prophet of PIT1 gene variation and its effect on growth traits in Chinese indigeous goat. J. Anim. Vet. Adv. 2010;9:2940–6. [Google Scholar]
• 43.Zeng X.C., Chen H.Y., Jia B., Zhao Z.S., Hui W.Q., Wang Z.B., et al. Identification of SNPs within the sheep PROP1 gene and their effects on wool traits. Mol. Biol. Rep. 2011;38:2723–8. doi: 10.1007/s11033-010-0416-4. [DOI] [PubMed] [Google Scholar]
• 44.Dasen J.S., Rosenfeld M.G. Signaling and transcriptional mechanisms in pituitary development. Annu. Rev. Neurosci. 2001;24:327–55. doi: 10.1146/annurev.neuro.24.1.327. [DOI] [PubMed] [Google Scholar]
• 45.Basoa E., Alfonsi C., Perez J.E., Cequea H. Karyotypes on the scallops Euvola ziczac and Nodipecten nodosus from the Gulf of Cariaco, Sucre State, Venezuela. Bol. Inst. Oceanogr. Venez. 2000;39:49–54. [Google Scholar]
• 46.Simakov O., Marletaz F., Cho S.J., et al. Insights into bilaterian evolution from three spiralian genomes. Nature. 2013;493:526–31. doi: 10.1038/nature11696. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 47.Heliconius Genome Consortium. Butterfly genome reveals promiscuous exchange of mimicry adaptations among species. Nature. 2012;487:94–8. doi: 10.1038/nature11041. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 48.Catchen J.M., Amores A., Hohenlohe P., Cresko W., Postlethwait J. Stacks: building and genotyping loci de novo from short-read sequences. G3 (Bethesda) 2011;1:171–82. doi: 10.1534/g3.111.000240. [DOI] [PMC free article] [PubMed] [Google Scholar]

Supplementary Materials