FOLIA BIOLOGICA ET GEOLOGICA 61/1, 17–23, LJUBLJANA 2020 GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS THROUGH GENOME-WIDE ALLELE FREQUENCY FINGERPRINTS GENETSKA KARAKTERIZACIJA VZORCEV AJDE Z ODTISI FREKVENCE ALELOV V GENOMU Michelle M. NAY1, Stephen L. BYRNE2, Eduardo A. PÉREZ3, Achim WALTER3, Bruno STUDER1 http://dx.doi.org/10.3986/fbg0063 ABSTRACT Genetic characterization of buckwheat accessions through genome-wide allele frequency fingerprints Genomics-assisted breeding of buckwheat (Fagopyrum esculentum Moench) depends on robust genotyping meth- ods. Genotyping by sequencing (GBS) has evolved as a flex- ible and cost-effective technique frequently used in plant breeding. Several GBS pipelines are available to genetically characterize single genotypes but these are not able to repre- sent the genetic diversity of buckwheat accessions that are maintained as genetically heterogeneous, open-pollinating populations. Here we report the development of a GBS pipe- line which, rather than reporting the state of bi-allelic single nucleotide polymorphisms (SNPs), resolves allele frequen- cies within populations on a genome-wide scale. These ge- nome-wide allele frequency fingerprints (GWAFFs) from 100 pooled individual plants per accession were found to be highly reproducible and revealed the genetic similarity of 20 different buckwheat accessions analysed in our study. The GWAFFs cannot only be used as an efficient tool to precisely describe buckwheat breeding material, they also offer new opportunities to investigate the genetic diversity between different buckwheat accessions and establish variant data- bases for key material. Furthermore, GWAFFs provide the opportunity to associate allele frequencies to phenotypic traits and quality parameters that are most reliably described on population level. This is the key to practically implement powerful genomics-assisted breeding concepts such as marker-assisted selection and genomic selection in future breeding schemes of allogamous buckwheat. Key words: Buckwheat (Fagopyrum esculentum Mo- ench), genotyping by sequencing (GBS), population genom- ics, genome-wide allele frequency fingerprints (GWAFFs) IZVLEČEK Genetska karakterizacija vzorcev ajde z odtisi frekvence alelov v genomu Genomsko podprto žlahtnjenje ajde (Fagopyrum escu- lentum Moench) je odvisno od robustnih metod genotip- iziranja. Genotipiziranje s spremljanjem sekvenc (genotyp- ing by sequencing, GBS) se je razvilo kot fleksibilna in razm- eroma poceni metoda, ki se jo uporablja pri žlahtnjenju ras- tlin. Uporabnih je več virov GBS za genetsko karakterizacijo posamičnih genotipov, toda te metode niso primerne za predstavitev genetske raznolikosti vzorcev ajde, ki jih vzdržujemo v heterozigotni obliki, kar velja za odprto op- lodne populacije. Tu poročamo o razvoju GBS metode, ki, namesto prikazovanja bi-alelnega polimorfizma posa- meznih nukleotidov (single nucleotide polymorphisms, SNPs), pokaže frekvence alelov v populaciji na nivoju geno- ma. Ta prikaz frekvence alelov na nivoju genoma (genome- wide allele frequency fingerprints, GWAFFs) z združenimi sto posameznimi rastlinami vsakega vzorca se je pokazal kot visoko ponovljiv in je prikazal genetsko podobnost 20 različnih vzorcev ajde, ki smo jih analizirali v naši raziskavi. Metoda GWAFFs ni uporabna samo kot učinkovito orodje za natančen opis materiala za žlahtnjenje ajde, ponuja tudi možnosti raziskave genetskih razlik med različnimi vzorci ajde in omogoča zbirke podatkov. Nadalje, metoda GWAFFs omogoča povezovanje frekvenc alelov s fenotipskimi last- nostmi in kvalitativnih parametrov, ki so najbolj zanesljivo opisani na nivoju populacij. To je ključ za praktično uporabo z genomiko podprtega žlahtnjenja, kot je z genskimi mark- erji podprta selekcija in genomska selekcija z GWAFFs. Ključne besede: ajda (Fagopyrum esculentum Moench), genotipizacija s sekvenciranjem (GBS), populacijska ge- nomika, GWAFFs 1 Molecular Plant Breeding, Institute of Agricultural Sciences, ETH Zurich, Universitaetstrasse 2, 8092 Zurich, Switzerland, Bruno.studer@usys.ethz.ch 2 Teagasc Crop Science Department, Oak Park, Carlow, R93XE12, Ireland 3 Crop Science, Institute of Agricultural Sciences, ETH Zurich, Universitaetstrasse 2, 8092 Zurich, Switzerland NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 18 FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 Plant breeding plays a key role in order to meet the increasing demand for food of the world’s growing population. Sophisticated breeding strategies have been developed depending on the reproductive strate- gy of crop species, but they are all based on one com- mon principle: Through initial crossings, genes and alleles are reshuffled and among the hundreds or thousands of variants produced in the cross, variants outperforming the parents or combining desirable characteristics may be chosen (Fehr 1987; Stoskopf et al. 1999). To select the best variants, plant breeding depended for centuries on the trained breeder’s eye. With the establishment of genomics-assisted plant breeding techniques, breeders were given an aid to support the selection process. Genomics-assisted plant breeding incorporates knowledge of the genetic deter- minant of the trait of interest and allows to select supe- rior variants based on genetic data (Moose & Mumm 2008). This technique allows to select for multiple traits and usually within a fraction of time needed to measure them. Through the steep price drop for next- generation sequencing in the last decades (Wetter- strand 2019), genomics-assisted breeding became not only feasible for major crops but is becoming increas- ingly popular in orphan crops. The orphan crop buckwheat (Fagopyrum esculen- tum Moench) is a desired food crop since its gluten- free grains are high in antioxidants and essential amino acids (Li & Zhang 2001). Agronomically, buck- wheat has shown beneficial effects in crop rotations and is an attractive bee crop (Falquet et et al. 2015; Teboh & Franzen 2011), but its low seed-set, indeter- minate growth and susceptibility to abiotic stresses have hindered wide adoption of buckwheat as a cash- crop in Europe (Lichtenhahn & Dierauer 2000). Buckwheat breeding is conducted in several programs around the world, but is complicated by the heterosty- lous self-incompatibility system (Ueno et al. 2016). Several attempts have been made to transfer the self- compatibility of its sister species Fagopyrum homotro- picum Ohnishi (Matsui et al. 2003), while this was successful, the resulting lines often suffered from in- breeding depression (Campbell 1997). Hence, most buckwheat grown today is of outcrossing nature and accessions are maintained as diverse populations. This renders it difficult to fix beneficial alleles in the popu- lation, because the ‘superior’ plants selected at harvest time are already pollinated by the ‘inferior’ neighbour plants. Genomics-assisted breeding offers opportunities to select plants containing beneficial alleles based on ge- netic data and known marker-trait associations. A re- quirement for this are reliable genotypic data of breed- ing germplasm. For buckwheat with a haploid genome size of around 1.3 Gb (2n=16), several genomic resources have become available recently (Logacheva et al. 2011; Mizuno & Yasui 2019; Nagano et al. 2000; Yasui et al. 2016). A widely used genotyping method that has evolved as highly flexible is genotyping by sequencing (GBS) (Elshire et al. 2011). In the GBS workflow, genomic DNA is cut by a restriction enzyme and se- quencing adaptors are ligated to the cutting sites. After a PCR multiplication step, the fragments are short-read sequenced, which results in repeated coverage of thou- sands of genetic loci (Elshire et al. 2011). Through alignment of the sequencing reads to a reference ge- nome, a genotyping matrix can be derived that allows for further downstream analysis to compare genotypes or conduct genetic studies. Since buckwheat accessions are populations of genetically distinct individuals, standard genotyping and variant calling pipelines tai- lored for genotyping single individuals or inbred lines (Li et al. 2009; Mckenna et al. 2010), are of limited use. As an alternative, a large number of single plants can be genotyped and instead of determining the allele present at a certain genetic location, allele frequencies can be calculated. A shortcut and budget-friendly option repre- sents the pooling of multiple individuals before sequenc- ing, which proved to be a highly accurate method in perennial ryegrass (Lolium perenne) (Byrne et al. 2013). The main objective of this study was to find a reli- able genotyping method for detailed genetic analyses of buckwheat accessions. Specifically, we adapted the GBS and analysis protocol reported by Byrne et al. (2013) to calculate genome-wide allele frequency fin- gerprints (GWAFFs) and tested their accuracy in a rep- licated set of twenty diverse accessions. 1 INTRODUCTION NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 19FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 2.1 Plant material and DNA extraction Twenty accessions from Austria (AT), France (FR), Germany (DE), Russia (RU), Slovenia (SI), Ukraine (UA), Czech Republic (CZ) and Switzerland (CH) were grown with a sowing density of 180 seeds/m2 in field plots of 3 x 4m at the ETH Research Station for Plant Sciences in Lindau-Eschikon, Switzerland (47.449N, 8.682E, 520 m a.s.l.). The following accessions were used: Bamby (AT), Billy (AT), Buchsa (CH), Carolin (FR), Carte Noir (FR), Darja (SI), Devyatka (RU), Dialog (RU), Dikul (RU), Drollet (F), Kerntner Hadn (AT), La Harpe (F), Lileja (UA), MinI (DE), Orphe (F), Pyra (CZ), Rosa (CH), Temp (RU), Theophani (DE), Tussi (DE). DNA was extracted from leaf material cut out with an apple corer to ensure tissues are of approxi- mately the same size. For each accession, pooled sam- ples of 100 randomly selected plants were taken in triplicate. The plant material was f lash frozen in liq- uid nitrogen and milled using mortar and pistil. DNA was extracted using the DNeasy Plant Mini Kit (Qui- agen, Hilden, DE) according to the manufacturer rec- ommendation. 2.2 GBS library preparation and DNA sequen- cing GBS library was prepared with the restriction enzyme combination PstI and ApeKI at LGC (LGC Ltd, Ted- dington, UK) according to their in house protocol (Arvidsson et al. 2016). The libraries were 150bp sin- gle-end sequenced on an illumina HiSeq (Illumina, Inc., San Diego CA, USA) machine at a depth of ap- proximately 400 million reads for 60 samples. 2.3 Genome-wide allele frequency fingerprints Demultiplexed fastq files of the GBS data were used for the analysis. The reads were mapped to the avail- able genome assembly FES_r1.0 (Yasui et al. 2016) using BWA (Li & Durbin 2010). A single nucleotide polymorphism (SNP) database was developed by combining read data of the three sample replicates of each accession. At each site where the minimum read depth (RD) of 30 was achieved, the allele frequency of the variant allele was calculated for each cultivar and sites where more than 25 percent of samples had missing data (RD < 30) were removed. The average variant allele frequency at each site was determined [frequency of variant allele / (frequency of reference allele + frequency of variant allele)] and used to filter out sites where the average minor allele frequency was less than 0.01. Allele frequencies were also deter- mined for each sample replicate and a reduced SNP database was generated that included only variant sites with a minimum read depth of at least 30 in all samples. This dataset was used to analyse the similar- ity of replicates and accessions with R (R CORE TEAM 2008) using the libraries ‘psych’ and ‘pheat- map’. To calculate homozygosity and similarity be- tween accessions, the mean GWAFFs over the three replicates was calculated and loci were considered homozygous if the allele frequency was larger than 0.975 or lower than 0.025. 2 MATERIAL AND METHODS 3 RESULTS 3.1 Allele frequency calling and distribution Genotyping by sequencing of the 20 buckwheat acces- sions in triplicate resulted in 3.5-9.5 million reads per pooled sample. Mapping them to the available draft genome (Yasui et al. 2016) resulted in a database con- taining 40,696 SNPs after filtering. Calculation of the allele frequencies for each sample separately revealed 15,726 high-quality loci after filtering. These loci were distributed on 3363 out of the 387,594 scaffolds report- ed in the draft genome sequence. 3.2 Reproducibility of GWAFFs Replicated sampling of the populations resulted in highly comparable GWAFFs within the replicates (Fig- ure 1). The Pearson correlation between replicates of the same accession ranged between 0.971 and 0.999, while between pooled samples of different accessions it ranged between 0.320 and 0.983. All but the accessions Tussi, Theophanu and MinI showed an allele frequen- cy distribution skewed towards the right, indicating that the alternative alleles were present at a low fre- quency. For Tussi, Theophanu and MinI the allele fre- quencies were distributed around the extremes (1 or 0, NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 20 FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 data for Tussi shown in figure 1), implying that the ac- cessions were highly homozygous, which is a conse- quence of their self-pollinating reproduction system transferred from F. homotropicum (F.J. Zeller, personal communication). Figure 1: Correlation matrix of genome-wide allele frequency fingerprints within and between the replicated pooled samples of the accessions Carte Noir, Devyatka and Tussi. In the upper diagonal, Pearson correlations between the samples are shown. In the diagonal, histograms of the allele frequency distribution for each sample are shown. In the lower diagonal, the allele fre- quencies of the two samples are plotted against each other with the red line representing the LOESS (locally estimated scatter- plot smoothing) line. Slika 1: Korelacijska matrica frekvenc odtisov v genomu v in med združenimi vzorci akcesij Carte Noir, Devyatka in Tussi. V zgornji diagonali so prikazane Pearsonove korelacije med vzorci. V diagonali so prikazani histogrami razporeditve frekvenc alelov za vsak vzorec. Pod diagonal so prikazane frekvence alelov dveh vzorcev, rdeča črta označuje LOESS (locally estimated scatterplot smoothing) linijo. 3.3 Homozygosity within buckwheat accessions The buckwheat accessions showed little homozygosi- ty, with the exception of the self-compatible acces- NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 21FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 sions Tussi, Theophanu and Min1 (Table 1). The range of heterozygosity excluding the self-compatible lines ranged from 66.7% (Carte Noir) to 88.0% (Darja) with a mean value of 80.3%. Accession Homozygosity [%] Bamby 22.7 Billy 13.4 Buchsa 17.7 Carolin 17.7 Carte Noir 33.3 Darja 12.0 Devyatka 19.4 Dialog 23.4 Dikul 18.5 Drollet 23.5 Table 1: Homozygosity rate of 20 buckwheat accessions based on genotyping by sequencing data of 15,726 genetic loci. Genetic loci were regarded as homozygous, if the allele frequency was higher than 97.5% or lower than 2.5%. Razpredelnica 1: Stopnje homozigotnosti 20 vzorcev ajde, zasnovane na genotipizaciji s pomočjo sekvenciranja 15.726 genetskih lokusov. Lokusi so bili upoštevani kot homozigotni, če je bila pogostnost alela višja od 97,5 % ali nižja od 2,5 %. Figure 2. Correlation analysis of genome-wide allele frequency fingerprints of 20 buckwheat accessions. Pairwise Pearson correla- tions were calculated using the mean allele frequency of the three replicates per accession. On the left side, a hierarchical cluster- ing analysis of the cultivars based on the correlation matrix is shown. Slika 2. Korelacijska analiza odtisov alelnih frekvenc po celotnem genomu za 20 akcesij ajde. Pearsonove korelacije so izračunane po parih z uporabo srednjih frekvenc alelov treh ponovitev na vsako akcesijo. Na levi strain je prikazano hierarhično združevanje podatkov analiz kultivarjev, zasnovano na korelacijski matrici. Accession Homozygosity [%] Kerntner Hadn 22.3 La Harpe 29.7 Lileja 15.4 MinI 91.8 Orphe 15.5 Pyra 15.4 Rosa 16.6 Temp 25.9 Theophanu 79.0 Tussi 85.7 3.4 Genetic similarity of accessions The genetic similarity of the accessions was analysed based on a correlation analysis of the mean GWAFFs NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 22 FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 of the 20 accessions (Figure 2). The homozygous, self- compatible accessions Tussi, Theophanu and MinI clustered separately and were distinct from the re- maining accessions with Pearson correlations of 0.33- 0.54. Within the heterozygous, self-incompatible ac- cessions, several were found to be highly similar, often clustering by the country of origin. A high genetic similarity was revealed between the Central European accessions Bamby, Kerntner Hadn, Darja, Buchsa, Orphe, Rosa, Lileja and Pyra, the Russian accessions Dikul, Devyatka, Dialog and Temp, and the French ac- cessions Carte Noir, La Harpe, Carolin and Drollet (Figure 2). 4. DISCUSSION 4.1 Genome-wide allele frequency fingerprints allow precise genotyping of buckwheat accessions In this study, we have shown that by pooling 100 indi- vidual plants of a buckwheat accession and subjecting them to GBS, representative and highly reproducible GWAFFs can be obtained. This allowed us to geneti- cally characterize 20 buckwheat accessions at unprec- edented precision. We identified genetically similar accessions and found that they often cluster by the re- gion of their origin. The genetic similarity of acces- sions bred in the same country or region may be an indication of the narrow genetic base in each country with limited gene-flow between breeding programs. Analysis of further accessions would yield a better un- derstanding of the buckwheat genetic resources world- wide and may allow to set exchange and conservation priorities. 4.2 Importance of robust genotyping method to implement genomics-assisted breeding in buck- wheat Accurate genotypic data, such as the GWAFFs present- ed in this work, are crucial to describe buckwheat breeding materials and investigate the genetic diversity between different buckwheat accessions. Furthermore, they enable to associate allele frequencies to plant phe- notypic traits and nutrition quality parameters that are most reliably obtained for accessions rather than single plants. In our study the sequencing reads were aligned to the publicly available reference genome (Yasui et al. 2016), which is still fragmented and does not assign chromosome numbers to the scaffolds. With the up- coming high-quality assembly by NRGene (NRGENE 2018), a better understanding of the genetic distances between polymorphisms and their density on the chromosome will be possible. Assigning genomic loca- tions to the polymorphisms genotyped will increase the efficiency to select for superior germplasm in fu- ture crossing-experiments via marker-assisted or genomic selection, and may allow to find candidate genes for certain traits. 4.3 High heterozygosity within buckwheat ac- cessions emphasizes the need to use population genetics approaches This study was the first to genetically describe buck- wheat materials using allele frequencies instead of bi- allelic SNPs. We found that on average 80.3% of the genetic loci covered by GBS were heterzygous. Hence, genotyping a single plant to represent the entire ge- nepool of an accession would result in missing out a large part of the diversity. The accession-specific allele frequencies can, however, be dynamic; for example ge- netic drift may act if population sizes are small (e.g. seed multiplication from a small batch of seeds) or if certain genotypes within the population cope better with new climatic conditions and therefore contribute more seeds to the next generation (Wright 1931). How these dynamics have affected buckwheat populations in the past is not known, but documenting the changes in allele frequencies in the future may allow to better understand the genetic basis of adaption to new envi- ronmental conditions (Günther & Coop 2013). NAY, BYRNE, PÉREZ, WALTER, STUDER: GENETIC CHARACTERIZATION OF BUCKWHEAT ACCESSIONS 23FOLIA BIOLOGICA ET GEOLOGICA 61/1 – 2020 5. REFERENCES Arvidsson, S., Fartmann, B., Winkler, S. & Zimmermann, W., 2016: Efficient high-throughput SNP discovery and genotyping using normalised Genotyping-by-Sequencing (nGBS). https://biosearch-cdn.azureedge.net/as- setsv6/efficient-high-throughput-snp-discovery-genotyping-ngbs-app-note.pdf (19.09.2019) Byrne, S., Czaban, A., Studer, B., Panitz, F., Bendixen, C. & Asp, T., 2013: Genome Wide Allele Frequency Fingerprints (GWAFFs) of Populations via Genotyping by Sequencing. PLoS ONE 8 (3). Elshire, R. J., Glaubitz, J. C., Sun, Q., Poland, J. A., Kawamoto, K., Buckler, E. S. & Mitchell, S. E., 2011: A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PLoS ONE 6 (5): 1–10. Falquet, B., Gfeller, A., Pourcelot, M., Tschuy, F. & Wirth, J., 2015: Weed suppression by common buck- wheat: A review. Environmental Control in Biology 53 (1): 1-6. Fehr, W. R. (1987): Principles of cultivar development. Theory and technique. New York. Günther, T. & Coop, G., 2013: Robust identification of local adaptation from allele frequencies. Genetics 195 (1): 205–220. Li, H. & Durbin, R., 2010: Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinfor- matics 26 (5): 589–595. Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G. & Durbin, R., 2009: The Sequence Alignment/Map format and SAMtools. Bioinformatics 25 (16): 2078–2079. Li, S. Q. & Zhang, Q. H., 2001: Advances in the development of functional foods from buckwheat. Critical Re- views in Food Science and Nutrition 41 (6): 451–464. Lichtenhahn, M. & Dierauer, H., 2000: Buchweizen. Frick. Logacheva, M. D., Kasianov, A. S., Vinogradov, D. V., Samigullin, T. H., Gelfand, M. S., Makeev, V. J. & Penin, A. A., 2011: De novo sequencing and characterization of floral transcriptome in two species of buck- wheat (Fagopyrum). BMC Genomics 12,30. Matsui, K., Tetsuka, T., Nishio, T. & Hara, T., 2003: Heteromorphic incompatibility retained in self-compatible plants produced by a cross between common and wild buckwheat. New Phytologist 159 (3): 701–708. Mckenna, A., Hanna, M., Banks, E., Sivachenko, A., Cibulskis, K., Kernytsky, A., Garimella, K., Altshu- ler, D., Gabriel, S., Daly, M. & Depristo, M. A., 2010: The Genome Analysis Toolkit: A MapReduce frame- work for analyzing next-generation DNA sequencing data. Genome Research 20 (9): 254–260. Mizuno, N. & Yasui, Y., 2019: Gene flow signature in the S-allele region of cultivated buckwheat. BMC Plant Bio- logy 19 (1): 1–9. Moose, S. P. & Mumm, R. H., 2008: Molecular Plant Breeding as the Foundation for 21st Century Crop Improve- ment. Plant Physiology 147 (3): 969–977. Nagano, M., Alp, J., Campbell, C. & Kawasakp, S., 2000: Genome size analysis of the genus Fagopyrum. Fago- pyrum 39:35–39. NRGENE, 2018: NRGene assembles first accurate buckwheat genome. https://www.nrgene.com/press-releases/ nrgene-assembles-buckwheat-genome/ (21.08.2019) R CORE TEAM, 2008: R: A language and environment for statistical computing. Stoskopf, N. C., Tomes, D. T. & Christie, B. R., 1999: Plant breeding: theory and practice. Jodhpur, India. Teboh, J. M. & Franzen, D. W., 2011: Buckwheat (Fagopyrum esculentum Moench) potential to contribute solu- bilized soil phosphorus to subsequent crops. Communications in Soil Science and Plant Analysis 42 (13): 1544–1550. Ueno, M., Yasui, Y., Aii, J., Matsui, K., Ota., T., Ueno, M., Yasui, Y., Aii, J., Matsui, K., Sato, S. & Ota, T., 2016: Genetic analyses of the heteromorphic self-incompatibility (S) locus in buckwheat. In: ZHOU, M., I, KREFT., WOO, S.-H., CHRUNGOO, N. & WIESLANDER, G. Hrsg.: Molecular Breeding and Nutritional Aspects of Buckwheat. 411–421. Wetterstrand, K. A., 2019: Data from the NHGRI Genome Sequencing Program (GSP). www.genome.gov/se- quencingcostsdata (19.08.2019) Wright, S., 1931: Evolution of mendelian populations. Genetics 16 (97): 97–159. Yasui, Y., Hirakawa, H., Ueno, M., Matsui, K., Katsube-Tanaka, T., Yang, S. J., Aii, J., Sato, S. & Mori, M., 2016: Assembly of the draft genome of buckwheat and its applications in identifying agronomically useful genes. DNA Research 23 (3): 215–224.