GONG Qian-chun, YU Hong-xiao, MAO Xin-rui, Ql Hui-dong, SHl Yan, XlANG Wei, CHEN Qing-shan,Ql Zhao-ming

Key Laboratory of Soybean Biology, Ministry of Education/College of Agriculture, Northeast Agricultural University, Harbin 150030,P.R.China

1. lntroduction

Soybean (Glycine max(L.) Merr), which is widely planted in the United States, Brazil, Argentina, China, and India(Thuet al.2014), contains high levels of proteins, linoleic acid, phospholipids, and a variety of human essential amino acids and plays a significant role in global food security(Medicet al. 2014). There are 18 types of amino acids in soybean, such as Met (methionine) (1.8 g/16 g nitrogen(N) on average), Lys (lysine) (6.1 g/16 g N on average),Cys (cysteine) (1.2 g/16 g N on average), Thr (threonine)(4.0 g/16 g N on average), and Ala (alanine) (4.6 g/16 g N on average) (Kwanyuen and Burton 2010), and amino acid composition also had a crucial influence on the nutritional value of soybean. Therefore, obtaining an optimal proportion of amino acids that meets food nutrition requirements should be considered in cultivar breeding (Qiuet al. 2014).

Soybean amino acid content is an important but complex quantitative trait controlled by multiple genes (Qiuet al.2014). Many soybean quantitative trait loci (QTLs) have been mapped in various genetic backgrounds and environments.Pantheeet al. (2006a) found four QTLs associated with Cys content and three QTLs related to Met content in soybean seeds. Warringtonet al. (2015) mapped two Lys-related QTLs to Gm08 and Gm20, and three QTLs for Thr to Gm01,Gm09, and Gm17. Wanget al. (2015) detected eight QTLs for both Cys and Met contents over three environments.However, the complicated mapping methods, large number,and large confidence intervals of these QTLs make it difficult to use them directly in breeding. Thus, these “original”QTLs need to be further screened to identify QTLs with high heritability and short confidence intervals (CI) (Chardonet al. 2004). The meta-analysis method was first proposed by psychologist Glass (1976) and has been widely applied in the fields of medical science, sociology, and behavioral science. To date, meta-analysis has been widely used in maize (Haoet al. 2010; Verretet al. 2017), rice (Trijatmikoet al. 2014), wheat (Wanget al. 2017), potato (Yellareddygariet al. 2016), and soybean (Thilakarathna and Raizada 2017). The first time that meta-analysis and the overview method were used to integrate crop QTLs was in a study by Chardonet al. (2004), who identified 62 “real” QTLs with small CIs based on the integration of 313 maize flowering time QTLs. Guoet al. (2006) were the first to apply metaanalysis to soybean. They identified 17 “real” QTLs for soybean cyst nematode resistance from 62 original QTLs.Compared with the large amount of data required for metaanalysis, overview analysis requires less data and yields more accurate results (Liuet al. 2011). The integration of meta-analysis and overview analysis has been widely used in further QTL analysis. Since Chardonet al. (2004),meta-analysis and/or overview methods have been used in studies of many soybean agronomic traits, including 100-seed weight (Sunet al. 2012b), plant height (Sunet al.2012a), oil content (Qiet al. 2011), protein content (Qiet al.2011), fungal disease resistance (Wanget al. 2010), insect resistance (Wanget al. 2009), phosphorus efficiency (Huanget al. 2011), and growth (Wuet al. 2009). However, to date,meta-analysis and overview analysis have rarely been used in studies of soybean amino acid QTLs. Using metaanalysis, Qiuet al.(2015)integrated 33 sulfur-containing amino acid QTLs based on the Consensus Map 4.0 high density genetic map and identified eight real QTLs (2015).In our study, the physical map was applied innovatively to meta-analysis because the genetic map was used most until now, and we combined meta-analysis with overview analysis to obtain more accurate amino acid content QTLs than previous studies (Liuet al. 2011; Gaoet al. 2013). A comprehensive map of 18 soybean amino acid contentrelated QTLs was constructed using the BioMercator ver.2.1 software, and the “real” large-effect QTLs with small CIs were obtained using meta-analysis and overview analysis.Furthermore, QTL candidate genes were annotated the Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Swissprot databases. Our study lays the foundation for further fine mapping and analysis of gene functions.

2. Materials and methods

2.1. Collection of soybean amino acid QTL information

The soybean amino acid-related QTLs in this study were collected from previously published papers (Table 1). We obtained 140 original QTLs for 18 different types of amino acids that were originally mapped from recombination inbred line (RIL) populations. In addition, basic information, such as mapping method, flanking markers, most likely position,95% confidence interval (CI), LOD value, meanR2, were collected for use in meta-analysis and overview analysis.

Table 1 Information of soybean amino acid original QTLs

2.2. QTL integration with a physical map

The Williams82 physical map obtained from Phytozome (https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Gmax) was used as a reference map. The Williams82 physical map could be easily used to integrate QTLs identified using different methods and genetic backgrounds. QTLs were projected from the original map onto the Williams82 physical map according to the most likely physical positions and CIs, and the projected QTLs were used to establish a consensus map using BioMercator ver. 2.1 software (Arcadeet al. 2004).

2.3. Meta-analysis of amino acid content QTLs

There were numerous QTL clusters in the consensus map,laying a solid foundation for searches of consensus QTLs.By analyzing each QTL cluster using the tools-Meta-analysis function, we obtained the most likely positions, 95% CIs and average contribution rate of the consensus QTLs. The meta-analysis function calculates the positions of consensus QTLs by integrating several independent QTLs on the same chromosome. This analysis yielded four models, and the optimal model called the “consensus QTL” was determined based on the minimum Akaike Information Criteria (AIC)(Gof finet and Gerber 2000). According to the maximum likelihood function ratio, each model specified the most likely position on the chromosome obeying Gauss’s law.The formula referred from Goffinet and Gerber (2000) was applied to calculate.

2.4. Overview analysis of amino acid content QTLs using a physical map

Normal distributions of the most likely QTL chromosomal locations were obtained from previously published papers.The most likely QTL position was calculated based on the CI (eq. (1)):

The normal function value of every position in 0.5 Mb steps fromxtox+0.5 was calculated using the function NORMDIST (Pi, QTL position;Si, false=0) in Excel.Pirefers to the most likely location of the chromosome.Si, variance,is calculated from eq. (1). QTL position refers to the genetic QTL location on the chromosome (Chardonet al. 2004).False=0 returns the probability density function. FunctionP(x) is set as the probability density function indicating the likelihood of a QTL being real in every experiment (eq.(2)).U(x), as a ruler ofP(x), realizes its function by estimating the unified possibility of QTL existence in every unit of chromosomes in single experiment.

In eq. (3), nbQTL represents the total number of QTLs,and nbE is the sum of the number of experiments where a QTL was identified on a chromosome. ThePivalues were plotted on the same graph to identify regions where there was a conspicuous peak in probability density to determine the location of “real” QTLs, chromosomal location was plotted on the horizontal axis, andP(x),U(x), andH(x) were plotted on the ordinate axis. The location of a “real” QTL was de fined as the peak value whereP(x) is greater thanU(x).

2.5. Gene mining from consensus QTL intervals

Candidate genes from consensus QTL regions were obtained from Phytozome and annotated using gene annotation databases, including GO (http://www.geneontology.org/),KEGG (http://www.genome.jp/kegg/), Swissprot (http://www.gpmaw.com/html/swiss-prot.html), NT (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastn&PAGE_TYPE=BlastSearch&LINK_LOC=blasthome),and NR (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastp&PAGE_TYPE=BlastSearch&LINK_LOC=blasthome).

3. Results

3.1. Analysis and integration of soybean amino acids amino acid-related QTLs

A total of 140 original QTLs related to 18 amino acids were obtained from literature searches (Appendix A). The numbers of the original amino acid-related QTLs were as follows: six Ala, one Arg (arginine), four Asp (aspartic acids),two Cys, eight Glu (glutamic acid), 61 Gly (glycine), one His(histidine), five Ile (isoleucine), seven Leu (leucine), three Lys, six Met, six Phe (phenylalanine), four Pro (proline), six Ser (serine),five Thr, six Trp (tryptophane), five Tyr (tyrosine), and four Val(valine). Two QTLs related to Gly haven’t been mapped to a specific chromosome, so a total of 138 QTLs were mapped onto the reference map. The maximumR2was 56%, and LOD score ranged from 2 to 8.8. All of the amino acid-related QTLs were mapped using a RIL population.

3.2. Meta-analysis of soybean amino acid-related QTLs

A total of 33 clusters of consensus QTLs were identified based on the chromosomal locations of the 138 original QTLs (Table 2). Each QTL cluster was analyzed using the tools-Meta-analysis in BioMercator ver. 2.1. The original QTLs mapped to Gm10 and Gm12 showed a dispersed distribution, and there was no overlap between them. As a result, there were no available consensus QTLs from these chromosomes.R2ranged from 0.2 to 32.7%. There were six consensus QTLs, located on chromosomes Gm04, Gm07,Gm09, Gm11, Gm14, and Gm18, withR2values greater than 20.0%. The QTL CIs were clearly reduced after metaanalysis, and the average CI decreased from 4.90 to 3.80 Mb.Six original QTLs on Gm01, Gm13, and Gm19 were combined into one consensus QTL and five on Gm02 and Gm14 were combined into one by meta-analysis. The CIs of nine consensus QTLs were less than 1.00 Mb. The coordinates of the left and right markers were 1.14–1.21 Mb on Gm07, 4.21–4.40 Mb on Gm07, 5.67–5.84 Mb on Gm08, 6.48–7.14 Mb on Gm09, 5.12–5.44 Mb on Gm13,8.21–9.22 Mb on Gm14, 0.55–1.12 Mb on Gm16, 2.82–3.45 Mb on Gm18, and 0.69–0.77 Mb on Gm19. The minimum CI was 0.07 Mb for a QTL on Gm07. What’s more, two of the nine QTLs with the shortest CIs and highest contribution rates, which were 20.3 and 21.0% for the QTLs on Gm07 and Gm14, respectively (Table 2).

3.3. Overview analysis of soybean amino acid-related QTLs

A total of 57 “real” QTL positions were mapped from the 138 original QTLs. The curve showed peaks on Gm08 and Gm18, where five QTLs were mapped respectively. There was one QTL each on chromosomes Gm02, Gm14, and Gm20. The number of QTLs on the other chromosomes ranged from two to four. We next used overview analysis, to further reduce the number of QTLs. Using overview analysis we integrated the original QTLs to obtain “real” QTLs. Using overview analysis, we obtained four chromosomes on which the number of QTLs reduced more than 6 from original QTLs respectively. The 10 original QTLs on Gm02 were reduced to one “real” QTL, the eight original QTLs on Gm07 werereduced to two “real” QTLs, the 11 original QTLs on Gm09 were reduced to four “real” QTLs, and the 15 original QTLs on Gm19 were reduced to three “real” QTLs (Table 3).

Table 2 Result of meta-analysis of soybean amino acids QTLs1)

3.4. Comparison between meta-analysis and overview analysis

QTLs identified using meta-analysis and overview analysis had similar CIs and chromosomal positions. Compared with meta-analysis, 24 more “real” QTLs were obtained with overview analysis. No “real” QTLs were identified on chromosomes Gm06, Gm07, Gm10, and Gm12 with either method. “Real” QTLs at 27 positions of 16 chromosomes were obtained using meta-analysis and overview analysis each separately. Compared with those obtained from metaanalysis, narrower CIs were obtained for QTLs located on Gm01 (44.50–45.50 Mb), Gm03 (37.00–39.00 Mb), Gm05(4.50–5.50 Mb), Gm09 (6.50–7.00 Mb and 43.00–44.00 Mb),Gm11 (8.00–9.00 Mb), Gm15 (19.50–27.50 Mb), Gm16 (0.50–1.00 Mb), Gm17 (10.00–11.00 Mb), Gm18 (3.00–3.50 Mb and 42.00–44.00 Mb), Gm19 (33.50–35.50 Mb) and Gm20(39.50–40.50) (Table 4). In Fig. 1, the positions and CIs of the QTLs on Gm03 and Gm09 were consistent between the two methods. A consensus QTL was mapped to 37 Mb on Gm03 using both meta-analysis and overview analysis.QTLs were mapped to coincident CIs on Gm09 for both meta-analysis and overview analysis (Table 4).

Table 3 QTLs position after overview analysis1)

Table 4 QTLs confidence intervals mapped by two methods

3.5. Gene mining from consensus QTLs

To identify QTL candidate genes, we screened 725 genes located within the CIs of the 33 consensus QTLs (Appendix B). More candidate genes are located in QTLs on Gm09(15.03–34.88 Mb), Gm02 (19.46–32.35 Mb), and Gm06(22.51–39.34 Mb) (119, 62, and 55 genes, respectively)than in the other QTLs. Only one gene is located within the QTLs located at 1.14–1.21 Mb on Gm07, 5.67–5.83 Mb on Gm08, and 5.11–5.44 Mb on Gm13, and there is no gene within 4.20–4.40 Mb on Gm07 (Table 5).

The total number of candidate genes with annotations was 725. Of these, 225 had GO annotations, 73 had KEGG annotations, 260 had Swissprot annotation, 568 had nr annotation, and 585 had nt annotation (Appendix C). Finally,20 candidate genes were identified, of which four genes were published (Appendix D).SDD1(Glyma.03G167600)encodes a subtilisin-like Ser protease (von Grollet al.2002).AtmBAC2(Glyma.04G143500) is a mitochondrial Arg transporter (Catoniet al. 2003). Threonine residues in the activation loop of the catalytic subdomain VIII ofAtSERK1(Glyma.05G083100) are potential targets for phosphorylation (Shahet al. 2001; Lewiset al. 2010).PYRB(Glyma.09G110200) controls the amino acid binding(Hooveret al. 1983). The 16 candidate genes which control the contents of Ser, Thr, Tyr, Lys, and Asp are not published(Table 6).

Fig. 1 QTL locus control chart of meta-analysis and overview analysis.

4. Discussion

4.1. Physical map

With the completion of genomic sequencing, high-density physical molecular maps with higher accuracy compared with genetic maps have been constructed. Physical maps describe the exact location of the known DNA markers,although the genomes are large, complex, and polyploid(Chenet al. 2010). Use of these maps also reduces the risk of missing data, simplifies the data processing steps,and expedites the experimental process. Schmutzet al.(2010) integrated physical and high-density genetic maps with the polyploid soybean genome sequence to create a chromosome-scale draft sequence assembly. The soybean physical map was applied innovatively in our study. Use of a high-density physical map as the reference map, solved the problem of projection when there was not a sufficient number of common markers between the genetic or physical map and the mapping population.

Table 5 Genes of all meta-analysis QTL sections

With the exploration of molecular makers, it will complement the Williams-82 reference genome (Liet al.2016) and will eventually bring new insights to both soybean research and soybean production.

4.2. Meta-analysis and overview analysis

Although the locations of QTLs can be determined efficiently based on the results of previous studies, the application of QTLs to crop improvement is still limited (Bolgeret al. 2014).The fact that the true chromosomal locations are unclear limits their use in breeding currently. It is important to have a complete view of a polygenic trait to optimize its use(Stocket al. 2016). Therefore, meta-analysis and overview analysis were used in this study to integrate published results. Overview analysis and meta-analysis have been used to obtain the physical QTL positions and accurate CIs (Gaoet al. 2013). The gene mapping could be greatly facilitated by the QTL meta-analysis and overview analysis(Dananet al. 2011a, b).

In recent years, meta-analysis and overview methods have been frequently used in soybeanin recent years. For example, meta-analysis and/or overview methods have been utilized to analyze QTLs for many soybean agronomic traits, such as oil content (Qiet al. 2011), protein content(Qiet al. 2011), plant height (Sunet al. 2012a), 100-seed weight (Sunet al. 2012b), fungal disease resistance (Wanget al. 2010), insect resistance (Wanget al. 2009), and phosphorus efficiency (Huanget al. 2011). Qiuet al. (2015)used software BioMercator ver. 2.1 to map 113 genes relatedto sulfur-containing amino acid enzymes and 33 QTLs controlling sulfur-containing amino acid content onto the Consensus Map 4.0, which was obtained by integrating the genetic and physical maps of soybean. Based on synteny between gene loci and QTLs and the QTL effect sizes, 16 candidate genes relating to the synthesis of sulfur-containing amino acids were identified (Qiuet al. 2014). In this study,33 consensus QTLs were obtained from 138 original QTLs using meta-analysis. When using another statistical method,overview analysis, the smaller CIs indicates that it is more accurate than meta-analysis. Qiuet al. (2014) previously used bioinformatic analysis of candidate gene copy number,SNP information and expression pro file to identify 12 genes involved in sulfur-containing amino acid metabolism on eight chromosomes. Compared with their study, our findings gave a more comprehensive picture of the loci involved in soybean amino acid metabolism. We identified 16 unpublished candidate genes controlling Ser, Thr, Tyr, Lys, and Asp contents based on GO annotation, as well as four candidate genes that have been identified by other studies (Table 6).In addition, some concrete gene functions and pathways were ascertained by gene annotation for the first time in this study. For instance, Glyma.05G091300 controls Thrtype endopeptidase activity within the proteasome pathway,and Glyma.17G157500 controls Asp-type endopeptidase activity in the Ribosome. Gene mining raises the potential for targeted interventions in soybean breeding.

Table 6 Candidate genes from meta-analysis results1)

Table 6 (Continued from preceding page)

5. Conclusion

Meta-analysis and overview analysis were conducted in this research. The consensus QTLs and “real” QTLs with good reproducibility were obtained from 138 original QTLs by two analysis methods. Thirty-three consensus QTLs were screened out by meta-analysis, of which the minimum CI was 0.07 Mb, the maximum average contribution rate was 32.7%. Fifty-seven “real” QTL positions were analyzed by overview analysis. And 16 unpublished candidate genes related with amino acids metabolism were identified. The results laid a foundation for fine mapping of soybean amino acid related QTLs and marker assisted selection.

Acknowledgements

This study was financially supported by the National Key R&D Program of China (2016YFD0100500, 2016YFD0100300,2016YFD0100201-21), the “Challenge Cup” National College Student Curricular Academic Science and Technology Works Competition of Ministry of Education of China (to Gong Qianchun, guided by Qi Zhaoming), the National Natural Science Foundation of China (31701449,31471516, 31401465, 31400074, 31501332), the China Post Doctoral Project (2015M581419), the Dongnongxuezhe Project (to Chen Qingshan), China, the Young Talent Project (to Qi Zhaoming) of Northeast Agriculture University,China (518062), the Heilongjiang Funds for Distinguished Young Scientists, China (JC2016004), and the Outstanding Academic Leaders Projects of Harbin, China(2015RQXXJ018).

Appendicesassociated with this paper can be available on http://www.ChinaAgriSci.com/V2/En/appendix.htm

Arcade A, Labourdette A, Falque M, Mangin B, Chardon F,Charcosset A, Joets J. 2004. BioMercator: Integrating genetic maps and QTL towards discovery of candidate genes.Bioinformatics, 20, 2324–2326.

Bolger M E, Weisshaar B, Scholz U, Stein N, Usadel B, Mayer K F. 2014. Plant genome sequencing-applications for crop improvement.Current Opinion in Biotechnology, 26, 31–37.

Catoni E, Desimone M, Hilpert M, Wipf D, Kunze R, Schneider A,Flügge U, Schumacher K, Frommer W B. 2003. Expression pattern of a nuclear encoded mitochondrial arginineornithine translocator gene fromArabidopsis.BMC Plant Biology, 3, 1.

Chardon F, Virlon B, Moreau L, Falque M, Joets J, Decousset L, Murigneux A, Charcosset A. 2004. Genetic architecture of flowering time in maize as inferred from quantitative trait loci meta-analysis and synteny conservation with the rice genome.Genetics, 168, 2169–2185.

Chen N W, Sévignac M, Thareau V, Magdelenat G, David P, Ash field T, Innes R W, Geffroy V. 2010. specific resistances againstPseudomonas syringaeeffectors AvrB and AvrRpm1 have evolved differently in common bean (Phaseolus vulgaris), soybean (Glycine max), andArabidopsis thaliana.New Phytologist, 187, 941–956.

Danan S, Veyrieras J B, Lefebvre V. 2011a. Construction of a potato consensus map and QTL meta-analysis offer new insights into the genetic architecture of late blight resistance and plant maturity traits.BMC Plant Biology, 11, 16.

Danan S, Veyrieras J B, Lefebvre V. 2011b. Genomic selection in tree breeding: Testing accuracy of prediction models including dominance effect.BMC Proceedings, 5,1753–6561.

Gao L F, Guo Y, Hao Z B, Qiu L J. 2013. Integration and“Overview” analysis of QTLs related to plant height in soybean.Hereditas, 35, 215–224.

Glass G V. 1976. Primary, secondary, and meta-analysis of research 1.Educational Researcher, 5, 3–8.

Gof finet B, Gerber S. 2000. Quantitative trait loci: A metaanalysis.Genetics, 155, 463–473.

von Groll U, Berger D, Altmann T. 2002. The subtilisin-like serine protease SDD1 mediates cell-to-cell signaling duringArabidopsisstomatal development.The Plant Cell, 14,1527–1539.

Guo B, Sleper D A, Lu P, Shannon J G, Nguyen H T, Arelli P R. 2006. QTLs associated with resistance to soybean cyst nematode in soybean: Meta-analysis of QTL locations.Crop Science, 46, 595–602.

Gutierrez-Gonzalez J J, Vuong T D, Zhong R, Yu O, Lee J D, Shannon G, Ellersieck M, Nguyen H T, Sleper D A.2011. Major locus and other novel additive and epistatic loci involved in modulation of isoflavone concentration in soybean seeds.Theoretical and Applied Genetics, 123,1375–1385.

Gutierrez-Gonzalez J J, Wu X, Gillman J D, Lee J D, Zhong R,Yu O, Shannon G, Ellersieck M, Nguyen H T, Sleper D A.2010. Intricate environment-modulated genetic networks control isoflavone accumulation in soybean seeds.BMC Plant Biology, 10, 105.

Gutierrez-Gonzalez J J, Wu X, Zhang J, Lee J D, Ellersieck M,Shannon J G, Yu O, Nguyen H T, Sleper D A. 2009. Genetic control of soybean seed isoflavone content: Importance of statistical model and epistasis in complex traits.Theoretical and Applied Genetics, 119, 1069–1083.

Hao Z, Li X, Liu X, Xie C, Li M, Zhang D, Zhang S. 2010.Meta-analysis of constitutive and adaptive QTL for drought tolerance in maize.Euphytica, 174, 165–177.

Hoover T A, Roof W D, Foltermann K F, O’Donovan G A,Bencini D A, Wild J R. 1983. Nucleotide sequence of the structural gene (pyrB) that encodes the catalytic polypeptide of aspartate transcarbamoylase ofEscherichia coli.Proceedings of the National Academy of Sciences of the United States of America, 80, 2462–2466.

Huang L L, Zhong K Z, Ma Q B, Nian H, Yang Y. 2011.Integrated QTLs map of phosphorus efficiency in soybean by Meta-analysis.Chinese Journal of Oil Crop Sciences,33, 25–32. (in Chinese)

Kassem M A, Meksem K, Iqbal M J, Njiti V N, Banz W J, Winters T A, Wood A, Lightfoot D A. 2004. Definition of soybean genomic regions that control seed phytoestrogen amounts.BioMed Research International, 2004, 52–60.

Kwanyuen P, Burton J W. 2010. A modi fied amino acid analysis using PITC derivatization for soybeans with accurate determination of cysteine and half-cystine.Journal of the American Oil Chemists’ Society, 87, 127–132.

Lewis M W, Leslie M E, Fulcher E H, Darnielle L, Healy P N, Youn J Y, Liljegren S J. 2010. The SERK1 receptorlike kinase regulates organ separation inArabidopsisflowers.The Plant Journal, 62, 817–828.

Li M W, Xin D, Gao Y, Li K P, Fan K, Muñoz N B, Yung W S, Lam H M. 2016. Using genomic information to improve soybean adaptability to climate change.Journal of Experimental Botany, 68, 1823–1834.

Liu S, Luo L, Liu Z X, Guan R X, Qiu L J. 2011. Integration of QTLs related to soybean protein content and “qualification”of them by overview method.Soybean Science, 30, 1–7.(in Chinese)

Medic J, Atkinson C, Hurburgh Jr C R. 2014. Current knowledge in soybean composition.Journal of the American Oil Chemists’ Society, 91, 363–384.

Panthee D R, Kwanyuen P, Sams C E, West D R, Saxton A M,Pantalone V R. 2004. Quantitative trait loci for β-conglycinin(7S) and glycinin (11S) fractions of soybean storage protein.Journal of the American Oil Chemists’ Society, 81,1005–1012.

Panthee D R, Pantalone V R, Sams C E, Saxton A M, West D R,Orf J H, Killam A S. 2006a. Quantitative trait loci controlling sulfur containing amino acids, methionine and cysteine, in soybean seeds.Theoretical and Applied Genetics, 112,546–553.

Panthee D R, Pantalone V R, Saxton A M, West D R, Sams C E. 2006b. Genomic regions associated with amino acid composition in soybean.Molecular Breeding, 17, 79–89.

Primomo V S, Poysa V, Ablett G R, Jackson C J, Gijzen M, Rajcan I. 2005. Mapping QTL for individual and total isoflavone content in soybean seeds.Crop Science, 45,2454–2464.

Qi Z M, Wu Q, Han X, Sun Y N, Du X Y, Liu C Y, Jiang H W, Hu G H, Chen Q S. 2011. Soybean oil content QTL mapping and integrating with meta-analysis method for mining genes.Euphytica, 179, 499–514.

Qiu H M, Li Z, Yu Y, Ma X P, Zheng Y H, Meng F F, Hou Y L,Wang Y Q, Wang S M. 2015. Mining and analysis of genes related to sulfur containing amino acids in soybean and bioinformatic analysis based on meta-QTL.Chinese Journal of Oil Crop Sciences, 37, 141–147. (in Chinese)

Qiu H M, Yang C M, Gao S Q, Hou Y L, Ma X P, Sun X M, Zheng H Y, Wang Y Q, Wang S M. 2014. Phenotype Identification and cluster analysis of soybean in upper latitudes regions based on protein content and amino acid composition.Journal of Plant Genetic Resources, 15, 1202–1208. (in Chinese)

Schmutz J I, Cannon S B, Schlueter J, Ma J, Mitros T, Nelson W, Hyten D L, Song Q, Thelen J J, Cheng J, Xu D, Hellsten U, May G D, Yu Y, Sakurai T, Umezawa T, Bhattacharyya M K, Sandhu D, Valliyodan B, Lindquist E,et al. 2010.Genome sequence of the palaeopolyploid soybean.Nature,463, 178–183.

Shah K, Vervoort J, de Vries S C. 2001. Role of threonines in theArabidopsis thalianasomatic embryogenesis receptor kinase 1 activation loop in phosphorylation.Journal of Biological Chemistry, 276, 41263–41269.

Smallwood C J. 2012. Detection of quantitative trait loci for marker-assisted selection of soybean isoflavone genistein.MSc thesis, University of Tennessee, Knoxville.

Stock K F, Jönsson L, Ricard A, Mark T. 2016. Genomic applications in horse breeding.Animal Frontiers, 6, 45–52.

Stombaugh S K, Orf J H, Jung H G, Chase K, Lark K G,Somers D A. 2004. Quantitative trait loci associated with cell wall polysaccharides in soybean seed.Crop Science,44, 2101–2106.

Sun Y N, Luan H H, Qi Z M, Shan D P, Liu C Y, Hu G H, Chen Q S. 2012a. Mapping and meta-analysis of height QTLs in soybean.Legume Genomics and Genetics, 3, 1–7.

Sun Y N, Pan J B, Shi X L, Du X Y, Wu Q, Qi Z M, Jiang H W, Xin D W, Liu C Y, Hu G H, Chen Q S. 2012b. Multi-environment mapping and meta-analysis of 100-seed weight in soybean.Molecular Biology Reports, 39, 9435–9443.

Thilakarathna M S, Raizada M N. 2017. A meta-analysis of the effectiveness of diverse rhizobia inoculants on soybean traits under field conditions.Soil Biology and Biochemistry,105, 177–196.

Thu N B A, Nguyen Q T, Hoang X L T, Thao N P, Tran L S P.2014. Evaluation of drought tolerance of the Vietnamese soybean cultivars provides potential resources for soybean production and genetic engineering.BioMed Research International, 2014,809736.

Trijatmiko K R, Prasetiyono J, Thomson M J, Cruz C M V, Moeljopawiro S, Pereira A. 2014. Meta-analysis of quantitative trait loci for grain yield and component traits under reproductive-stage drought stress in an upland rice population.Molecular Breeding, 34, 283–295.

Verret V, Gardarin A, Pelzer E, Médiène S, Makowski D,Valantin-Morison M. 2017. Can legume companion plants control weeds without decreasing crop yield? A metaanalysis.Field Crops Research, 204, 158–168.

Wang J, Song W K, Zhang W B, Liu C Y, Hu G H, Chen Q S.2009. Meta-analysis of insect-resistance QTLs in soybean.Hereditas, 31, 593–561.

Wang J L, Liu C Y, Wang J, Qi Z M, Hui L I, Hu G H, Chen Q S. 2010. An integrated QTL Map of fungal disease resistance in soybean (Glycine maxL. Merr): A method of meta-analysis for mining R genes.Agricultural Sciences in China, 9, 223–232.

Wang J Y, Xiong Y C, Li F M, Siddique K H, Turner N C. 2017.Effects of drought stress on morphophysiological traits,biochemical characteristics, yield, and yield components in different ploidy wheat: A meta-analysis.Advances in Agronomy, 143, 139–173.

Wang X, Jiang G L, Song Q, Cregan P B, Scott R A, Zhang J, Yen Y, Brown M. 2015. Quantitative trait locus analysis of seed sulfur-containing amino acids in two recombinant inbred line populations of soybean.Euphytica, 201,293–305.

Warrington C V, Abdel-Haleem H, Hyten D L, Cregan P B, Orf J H, Killam A S, Bajjalieh N, Li Z, Boerma H R. 2015. QTL for seed protein and amino acids in the Benning×Danbaekkong soybean population.Theoretical and Applied Genetics,128, 839–850.

Wu Q, Qi Z M, Liu C Y, Hu G H, Chen Q S. 2009. An integrated QTL map of growth stage in soybean [Glycine max(L.) Merr.]: Constructed through meta-analysis.Acta Agronomica Sinica, 35, 1418–1424. (in Chinese)

Yang K, Moon J K, Jeong N, Chun H K, Kang S T, Back K, Jeong S C. 2011. Novel major quantitative trait loci regulating the content of isoflavone in soybean seeds.Genes & Genomics,33, 685–692.

Yellareddygari S K R, Pasche J S, Taylor R J, Gudmestad N C. 2016. Individual participant data meta-analysis of foliar fungicides applied for potato early blight management.Plant Disease, 100, 200–206.

Yoshikawa T, Okumoto Y, Ogata D, Sayama T, Teraishi M, Terai M, Toda T, Yamada K, Yagasaki K, Yamada N, Tsukiyama T. 2010. Transgressive segregation of isoflavone contents under the control of four QTLs in a cross between distantly related soybean varieties.Breed Science, 60, 243–254.

Zeng G, Li D, Han Y, Teng W, Wang J, Qiu L, Li W. 2009.Identification of QTL underlying isoflavone contents in soybean seeds among multiple environments.Theoretical and Applied Genetics, 118, 1455–1463.