In MacadamiaGGD, we integrated the genetic information data, including that of five genomes of four species, the chloroplast and mitochondrion genomes of three species and genome annotations, which were previously released in public databases, including the National Center for Biotechnology Information (NCBI) Assembly database, the GigaScience database (GigaDB), and the China National Center for Bioinformation (CNCB) Genome Warehouse (GWH) database. In addition, transcriptomic data for three macadamia varieties were downloaded from the NCBI Sequence Read Archive (SRA) database. The germplasm, genetic linkage map, SNP and SSR marker data were retrieved from the NCBI PubMed database and other databases, as summarized in Table 1 . The components of data integration mainly include the data source, the data transform, and the data sink in the database. Extract, transform, and load (ETL) architecture was applied to data integration. In data integration process, raw data were collected, transformed, sorted, cleaned, aggregated, and stored via using PostgreSQL 9.5.25, Scala 2.13.1, AKKA 2.6.5, and SBT 1.3.5 ( Figures 1A, B ). Processed raw data were applied for variation calling and data visualization though using HTML5, CSS3, Java Script, Slick 3.3.2, Bootstrap 3.3.0 and Play Framework 2.8.2 ( Figure 1B ).

MacadamiaGGD was deployed in the Ubuntu 16.04 operation system using AKKA 2.6.5 (https://akka.io) as the web server, PostgreSQL 9.5.25 (https://www.postgresql.org) as the database server, Scala 2.13.1 (https://www.scala-lang.org) as the programming language and SBT 1.3.5 (https://www.scala-sbt.org) as the interactive building tool. All the data were managed and stored in the PostgreSQL Database. The website interface was generated via Bootstrap 3.3.0 (https://getbootstrap.com) and Play Framework 2.8.2 (https://www.playframework.com/). The web interface of MacadamiaGGD was developed using HTML5, CSS3, Java Script. The query function was enforced based on the Slick 3.3.2 middleware tier. JBrowse 1.16.6 (https://www.jbrowse.org) was used for genome visualization.

Leaf samples of 21 macadamia accessions for DNA isolation were collected from the macadamia plantation in Chongzuo, Guangxi, China ( Table S1 ). The DNA was isolated following a previously described method (Doyle, 1991), with slight modifications. To avoid problems of low efficiency and insufficient grinding due to manual grinding, young leaves were ground in a Tissuelyser-192 (Shanghai Jingxin Industrial Development Co., Ltd., China) and extracted with a 2% cetyltrimethylammonium bromide (CTAB) buffer. Nucleic acids were isolated with a chloroform: isoamyl alcohol (24:1) solution. DNA was purified with ethanol and resuspended in sterile distilled water. The DNA quality and concentration were assessed using ultraviolet spectrometry via a Nanodrop 2000c (Thermo Fisher Scientific, MA, USA) and agarose gel electrophoresis. The purified DNA was stored at -20°C until use.

New microsatellite markers were screened in the M. integrifolia HAES 741 reference genome (https://www.ncbi.nlm.nih.gov/bioproject/748012) by using SSRHunter 1.3 (http://www.bio2soft.net) (Li and Wan, 2005). The search criteria were set as 2, 3, and 4 nucleotides, corresponding to at least 4 repetitions. Afterward, the SSRs, comprising no fewer than 30 repeated motifs and being evenly distributed on each chromosome, were preferentially selected. To further confirm the quality of the SSR markers, each sequence was again queried via BLAST within MacadamiaGGD and tested via polymerase chain reaction (PCR).

Primer 3 (https://primer3.org) was used to design primer pairs flanking the sequences of the screened SSR motifs. The primer design parameters were as follows: primer length, 17-25 bp; melting temperature (Tm), 53 °C; amplicon size, 350-500 bp; and GC content, 40-60%.

The SSR PCR mixture (10 μL) comprised 1 μL of DNA, 0.4 μL of each primer (10 μM), 5 μL of Rapid Taq Master Mix (Vazyme, China) and 3.2 μL of double-distilled water. The amplification reaction program was as follows: 5 min at 95°C; 36 cycles of (30 s at 95°C, 53°C and 72°C); and a final extension of 5 min at 72 °C. Afterward, the mixture was held at 16°C. The PCR products were examined by electrophoresis on a 7% nondenaturing polyacrylamide gel run at 220 V for 40 min and visualized by silver staining. The density distribution map of polymorphic SSR markers on chromosomes was generated using MG2C software (http://mg2c.iask.in/mg2c_v2.1/).

MacadamiaGGD contains the most comprehensive bioinformatics datasets of macadamia (including five genomes, a total of 89.28 Gb of transcriptomic data, three chloroplast and mitochondrion genomes, germplasm data for four species and 262 main varieties, nine genetic linkage maps, 35 SNPs and 657 SSR markers), which provides convenient access to the large amount of germplasm and genomic information of macadamia ( Figure 1A ). MacadamiaGGD is composed of 11 main functional modules: Home, Germplasm, Genomes, Expression, BLAST, Markers, Maps, Tools, References, Download and Help ( Figure 1C ). MacadamiaGGD can be used to search and visualize genomic information by using various tools, including search, JBrowse, BLAST, primer designer, sequence fetch, enrichment analysis, multiple sequence alignment, genome alignment, and gene homology annotation ( Figure 1 ). MacadamiaGGD also provides information about macadamia germplasm and genome-related references. In summary, researchers can use the above functional modules of the database to quickly acquire the germplasm and genomic information of macadamia.

In the Germplasm module of MacadamiaGGD, 23 agronomic traits of four species and 16 agronomic traits of 262 main varieties were carefully described, including tree vigor, leaf type, fruit shape, flower color, the early-bloom stage and full-boom stage, and others. Users can easily obtain information on the morphological characteristics of four macadamia species and 262 varieties in the germplasm module. In addition, a phylogenetic analysis tool based on the results of Alam et al. (2018), which shows genetic distances between individuals genotypes, is provided in this module.

The MacadamiaGGD database provides public information on the assembled genomes of the M. integrifolia, M. tetraphylla, M. ternifolia, and M. jansenii, which are available in different public databases. For example, when “Genomes” is clicked on, the column header label appears, showing the suboptions as in Figure 2A . We can choose any label to access the sublinks and search for the needed information. When the user enters a gene “LOC122078696” in Macadamia integrifolia HAES 741 genome Browse, it will get the structure and function annotation information of all transcript of the gene ( Figure 2B ). Moreover, when the user clicks “Search”, a new layer appears with four options: “Keyword”, “Gene ID”, “Gene Name”, and “Region” ( Figure 2C ). Then, if one clicks “Gene ID”, the interface appears as a blank box ( Figure 2C ). The user can enter the gene “LOC122078696” in the box and click the Search button; then, the requested information is displayed ( Figure 2C ).

Gene annotations in MacadamiaGGD are displayed graphically in the genome JBrowse, which includes the information of the gene location, nucleotide sequences, amino acid sequences, and other features. For example, if a user selects the genomic region from 192751 bp to 203445 bp on Chromosome 14 (NC_056557.1) for browsing, all genes located within this zone are displayed properly ( Figure 2D ). Further, when the mRNA XM_042644823.1 is clicked on, detailed information on its mRNA, coding sequence (CDS), and other features are displayed ( Figure 2E ).

In the expression module of MacadamiaGGD, a total of 89.28 Gb of raw RNA-seq data were collected from tissues of young leaves, shoots, and flowers from the cultivar ‘Mauka’ (Nock et al., 2020); tissues of leaves, stems, flowers, and roots from the cultivar ‘Kau’; and shells and kernels at five different development stages from cultivar ‘Hinde’ (Lin et al., 2022). By mapping the transcriptome data to the reference genome and using transcripts per million (TPM) for calculation, we acquired the expression matrix of the annotated genes of macadamia.

BLAST is the most commonly used tool and is included as a separate module in the MacadamiaGGD database. It allows users to perform both BLASTp and BLASTn searches to rapidly align sequences to the database. In the BLAST module, pasting the DNA/protein sequences in the query box or uploading a FASTA file is acceptable. For example, the users can enter “Example 1” sequence in the blank box and select the against database type, e-value, and max target sequence number and then click the “Run” button to obtain the comparison results via the “BLASTp” function ( Figure 3A ). In addition, when pulling down the search result interface, a user is presented with all the comparison results ( Figure 3B ), including the description information of the candidate subject sequences alignment parameters ( Figure 3C ) and the matching information between the query sequence and each subject sequence ( Figure 3D ).

In the “markers” module, we included 657 SSR markers and 35 SNPs. Macadamia trees have a relatively long juvenile period (commonly four to five years); thus, it would take a great deal of time to select high-yielding cultivars for breeding. Molecular markers that are associated with key yield traits are extremely important for developing rapid cycle breeding programs in macadamia (O'Connor et al., 2020). To verify the polymorphism of SSR markers from previous research (Schmidt et al., 2006; Nock et al., 2014b; Langdon et al., 2019), we randomly selected 8 primer pairs from MacadamiaGGD ( Table S2 ) and identified polymorphisms of these SSRs via electrophoresis. The results showed that the selected primer pairs were polymorphic.

In this study, a total of 145593 SSR loci were obtained from M. integrifolia HAES 741 genomic sequences (Nock et al., 2020). They were evenly distributed on 14 chromosomes, with an average density of 10400 loci per chromosome ( Table 2 ). SSR motifs exist as one of three main types: dinucleotide repeats (DNRs), trinucleotide repeats (TNRs) and tetranucleotide repeats (TTRs). Among these SSRs, DNRs were the most abundant (115139), followed by TNRs (26400) and TTRs (4054), which accounted for 79%, 18% and 3%, respectively ( Table 2 ). A total of 927 primer pairs were designed by the selection of the SSR loci with repeat numbers ≥30 from the total SSR loci ( Table S3 ). Out of 927 amplified products, 605 primer pairs were polymorphic, with an average of 1.17 SSR markers per Mb on 14 chromosomes. According to the SSR density distribution map, chromosome 5 had the highest number of SSRs (81), but chromosome 12 had only 13 SSRs ( Figure 4 ). In addition, a total of 35 SNPs were included in the “markers” module, which were significantly associated with the yield component traits identified by genome-wide association studies (GWASs) (O'Connor et al., 2019a; O'Connor et al., 2020).

The map module contains nine genetic linkage maps derived from three macadamia cultivars, HAES 741, HVP A268 and HVP A4. In each map, there were 14 linkage groups (LGs), which correspond to the number of haploid chromosomes in macadamia. When the users open this module, the features of the maps are displayed, including the description and number of maps. The images of the maps appear at the lower left of the module, while the detailed information of the LG location, the marker numbers, the largest and smallest gap, the total length and the average length between markers is displayed at the lower right.

The tools module contains several utilities, including “Primer designer”, “Sequence Fetch”, “Enrichment analysis”, “Multiple sequence alignment”, “Genome alignment”, and “Gene homology annotation”, which allow a relatively complete bioinformatics analysis. The user can click the “Primer designer” button, input the nucleic acid sequence or select a scaffold range, adjust the appropriate parameters, and click the “Run” button to obtain a satisfactory pair of primers. Users can screen functional genes of interest (GOIs) based on the data of the M. integrifolia transcriptome, click the “Enrichment analysis” button, input the gene ID in the dialog box and select Kyoto Encyclopedia of Genes and Genomes (KEGG) or Gene Ontology (GO) for functional clustering analysis. “Sequence Fetch” can be used to efficiently obtain the sequence of GOI from the M. integrifolia genome, which can acquire either a certain or multiple gene sequences at the same time. “Muscle” is a multisequence alignment tool that not only can be used to obtain homology between genes but also can be used to build an intuitive diagram. The “primer designer” tool can be used to design specific primers to clone GOIs for functional research. In addition, by using the “LASTZ” and “GeneWise” tools, users can complete genome alignment and gene homology annotation, respectively.

Currently, the “Reference” module contains the macadamia germplasm and genome-related references, which allows users to query approximately 40 articles information related to the data contained in MacadamiaGGD. The completion and optimization of macadamia genome sequencing results among these publications contribute to the study of macadamia functional genomics and comparative genomics and are convenient for molecular plant breeding efforts.

MacadamiaGGD integrates BLAST, enrichment analysis, and other tools for functional genomic research of Macadamia. Acyltransferases are the potential molecular targets for genetic engineering to increase the oil content and alter the fatty acid composition in the oil crops (Zhang et al., 2021). Here, we provide a case study on the diacylglycerol acyltransferases (DGATs) of M. integrifolia by using the “BLAST”, “GO enrichment”, “JBrowse”, and “Gene Expression” function of MacadamiaGGD. By using BLAST in MacadamiaGGD, the Conserved Domains Database (CDD) of the NCBI database, the SMART database (https://smart.embl.de/) and MEGA 11 software (https://megasoftware.net/), we obtain one DGAT1 (MiDGAT1), three DGAT2 (MiDGAT2-1, MiDGAT2-2, MiDGAT2-3), and one DGAT3 (MiDGAT3) ( Figure 5A ). Of the five MiDGAT genes, two genes (MiDGAT2-1, MiDGAT2-3) were mapped to chromosome 14, and their physical positions were very close ( Figure 5B ).

To verify the expression features of MiDGATs during triacylglycerol (TAG) biosynthesis, we downloaded the transcriptome expression data of M. integrifolia kernel development from MacadamiaGGD. By using gene ontology (GO) annotation information available from MacadamiaGGD, we conducted the GO enrichment analysis of the five MiDGAT genes from M. integrifolia. The results showed the five MiDGATs were enriched in more than 30 GO terms, which are involved in fatty acid and TAG biosynthesis in plants ( Figure 5C ). Further, we also investigated the expression profile of MiDGATs at five stages of kernel development. MiDGAT2-1 and MiDGAT2-3 were highly expressed in stages I and II ( Figure 5D ). MiDGAT2-2 exhibited low expression levels in stages I and II, whereas it was highly expressed in stages III, IV, and V. Consistent with these results, The expression pattern of MiDGAT2 was recently found to be mainly correlated with fatty acid biosynthesis at different stages of developing kernels (Gao et al., 2021).