Evaluation of two high-throughput genotyping systems for rapid identification of Canadian wheat varieties

Abstract In this study, we report an updated panel of 32 DNA markers used for identification of wheat varieties and assess their performance in the OpenArray and SmartChip high-throughput genotyping systems. While both systems are unique and offer different advantages and disadvantages, both systems can successfully identify Canadian wheat varieties.


Introduction
Canadian wheat is marketed by class and designated varieties share similar class-specific functional characteristics and end-use quality. Currently, there are over 500 registered Canadian wheat varieties, which are subdivided into over 15 different classes. Canada Western Red Spring is the most widely grown wheat class in Canada, with AC Brandon being the most predominantly grown variety (CGC 2021). However, other wheat classes or varieties may be better suited to some growers, depending on their local climate, soil conditions, and on-farm management practices. Each year, new wheat varieties are also registered, have their registration canceled, or can change class. Testing the wheat variety composition during grain transactions is critical to support Canada's wheat grain quality assurance program to ensure that grain will have the desired properties for processing and making food products. It is also important in maintaining Canada's reputation as a preferred and reliable supplier of wheat. The Canadian grain industry requires reliable and high-throughput tools to identify wheat variety profiles during grain transactions, and DNA testing has emerged as the most efficient tool for wheat variety identification (Perry and Lee 2015).
Tools that detect DNA differences are favorable methods for variety identification in grain industries (Perry 2004;Banyai et al. 2006;Perry andLee 2015, 2016a). Reductions in the cost of DNA sequencing and increases in throughput have led to the release of accurate reference genomes that can be used for resequencing and the rapid detection of single nucleotide polymorphisms (SNPs) (Walkowiak et al. 2020). Individual SNPs have also become very easy to score and because they are often bi-allelic, both alleles can be assessed in a single assay to improve accuracy. Kompetitive allele-specific PCR (KASP, LGC Biosearch Tech), rhAmp SNP assay (Integrated DNA Tech), and TaqMan SNP genotyping assay (ThermoFisher  (Yoon et al. 2007) and the set of 7 SNPs from hops (Henning et al. 2015) have been tested for variety identification. Gao et al. (2016) developed a diagnostic panel of 43 SNPs that can identify 429 common wheat varieties in China. Similarly, Ndjiondjop et al. (2018) and Nguyen et al. (2020) investigated core sets of SNPs for identification of rice and pumpkin varieties, respectively. In our previous work, we selected 16 SNPs to identify 47 registered Canadian tetraploid or durum wheat varieties (Perry and Lee 2016b), and 32 SNPs to identify hexaploid or bread/noodle wheat varieties using the OpenArray system (Perry and Lee 2015).
Each crop type will require a different number of DNA markers depending on the genetic diversity of the crop, availability of DNA markers, and number of varieties that need to be distinguished. For Canadian hexaploid wheat, there are >511 registered varieties and 125 nonregistered or deregistered varieties, which we were able to distinguish using a panel of 32 assays and the OpenArray platform (Table S1). These varieties span 18 different wheat classes and distinguishing them is important for detection of contamination of grain samples with wheat of other classes, which vary in their quality profiles. The 32-marker panel is able to distinguish most varieties by two or more markers, making identification accuracy very high. The SmartChip genotyping sys-tem is a similar system to the OpenArray platform and also has excellent high-throughput capacity and the flexibility to use small sets of SNPs, such as 12, 24, 36, and 48 SNP formats. In this study, we have updated our 32 wheat SNP assays used in the OpenArray plate (Perry and Lee 2015) with improved markers that have greater accuracy and discriminatory power and have adapted these markers to the SmartChip system and investigated the accuracy and efficiency of both genotyping platforms for identifying Canadian wheat varieties.

Materials and methods
DNA samples were prepared from individual seeds in a 96-well format following a procedure described previously (Perry and Lee 2015). The seed samples were randomly selected from submitted samples of the cargo monitoring program of the Canadian Grain Commission in 2021. Thirty-two SNP markers that were previously validated in the OpenArray platform were selected for evaluation in the SmartChip system. These assays were slightly modified from our previously reported DNA marker panel with adjustments that increase the performance and accuracy of selected assays. The assays 2B-16 (#9 old) and 7B-02 (#31 old) were replaced by BE-1B-04 (#9) and 2B-31b (#19), which offered a better resolution for the separation of closely related lines such as AAC Brandon and AAC Elie, or CDC Landmark and CDC Hughes, respectively. There was also a relocation between #19 and #31 on the OpenArray chip, and modifications on the genomespecific primers of 3A-06 (#11), 3B-05-2 (#15), and 6A-06 (#25) ( Table S2).
The extracted DNA samples and the SNP markers were loaded into the SmartChip using the MultiSample NanoDispenser (Takara Bio Inc.) with the 36 assays × 144 DNA samples format and combined with the SmartChip qPCR master mix (640209, Takara Bio Inc.), as per the manufacturer's instructions (Fig. 1A). For the OpenArray, the Autoloader (Ther-moFisher Sci.) was used to load the 96 DNA samples into the OpenArray plate that was preloaded (preprinted) with the 32 markers (Fig. 1B). The loaded SmartChip and the Ope-nArray plate were amplified on the SmartChip cycler (Takara Bio Inc.) and the QuantStudio 12K Flex Real-Time PCR system (ThermoFisher Sci.), respectively. The cycling condition of the OpenArray plate was pre-set in the QuantStudio 12K Flex Real-Time PCR system as single default condition, and the SmartChip amplification consisted of an initial incubation for 9.75 min at 95 • C and followed by 40 cycles of 95 • C for 18 s and 60 • C for 1 min. After the amplification cycling, the fluorescence data for the VIC and FAM DNA dyes were exported as Microsoft Excel format ( * .xlsx) from the QuantStudio 12K Flex software and the SmartChip qPCR software. The exported data were then analyzed by the VID Inspector software (Perry and Lee 2015). For setting the boundaries of VIC and FAM genotyping groups, we applied minimum and maximum (for outlier) fluorescence values and acceptable slopes (the ratio of FAM/VIC) to each group. If a data point did not classify into either the VIC "1" or FAM "2" groups, it was considered as a missing data point (no call).

Results
Both the SmartChip and OpenArray platforms offer highthroughput and cost-effective low volume assay formats. The SmartChip has 5184 nanowells in 4 cm × 4 cm chip and individual reaction volumes are 100 nL (Fig. S1A). The OpenArray plate has 3072 holes in a 2 cm × 7 cm metal plate and has 33 nL reaction volumes (Fig. S1B). The side-by-side comparison of the 32 TaqMan SNP genotype assays previously validated in the OpenArray system showed excellent compatibility with the SmartChip system ( Figs. 1 and S2).
For some assays, the assignment of slopes, minimums, and maximums to distinguish the genotyping calls was simpler and resulted in more accurate genotyping calls for the SmartChip plots than the OpenArray plots (Fig. 1). This is particularly evident for the marker #6 2B-04 (Fig. 1A) and #19 2B-31 (Fig. 1C), where the genotype clusters for VIC and FAM were less distinct for the OpenArray plots; if more stringent rules were applied to these plots to increase accuracy, fewer data points would be assigned to clusters, resulting in an increased number of "no calls". The marker #20 4B-02 worked well in both systems, but there were more samples with "no call" (Fig. 1D, black dots) in the OpenArray plot. The marker #8 1B-17 (Fig. 1B) had an abundance of samples with "no call" in the OpenArray plot. This marker is located at a position on the array that is expected to have variable performance. In the OpenArray format, 8 out of 32 positions may be impacted by a loading feature of the system that causes cross-linking (bridging reaction holes) among the holes; the manufacturer therefore indicates that the 32 formats reliably support fewer assays. Nevertheless, these challenging locations on the plate can still provide valuable data. To compare the number of missed calls between the OpenArray and SmartChip systems, we considered the 32 and 24 assay configurations with and without these challenging locations on the plate (Table 1).
Overall, the SmartChip system produced few "no calls", 0.9% considering the 32-marker configuration and 1.0% considering the 24-marker configuration. In contrast, the Ope-nArray system produced 5.5% and 4.0% "no calls", respectively. In most cases, a few random missing data points will not affect the ability to identify a variety because the identification is based on several other markers in the panel. However, one mismatch can cause unidentified or different variety calls in a few sets of variety pairs and small groups. This can be a particular issue for genotypes that are derived from the same or similar parents, such as AAC Brandon and AAC Elie. For example, the marker #19 (Fig. 1C) is a critical marker distinguishing AAC Brandon and AAC Elie. In the OpenArray, four DNA samples failed to score marker #19; therefore, they resulted in an ambiguous call between the two varieties. However, in the SmartChip, those four samples were successfully genotyped, and the DNA samples were able to call the variety correctly. In our observation, single mismatches could mostly come from simple amplification error or unidentified variant(s) in the same variety. On average, the successful calls (per DNA sample) for the 32 markers was comparable at 31.7 for SmartChip and 30.3 for the OpenArray.
In our SNP marker panel, we have tried to select markers such that a minimum of three markers are distinguish- Fig. 1. Four genotyping plots of marker numbers 6,8,19,and 20 from 288 Canadian wheat DNA samples (blue: FAM group,red: VIC group,and black: no call). Green dotted lines are minimum florescent values and acceptable slope of FAM/VIC that set the boundaries of the two groups. The plots were generated by VID inspector software from the exported fluorescence data of OpenArray and the SmartChip systems using the same DNA samples. ing each variety; however, there are still some challenging pairs or small groups of varieties that are distinguished by less than three markers. In the OpenArray system, there is also increased likelihood of "no call", particularly at locations on the plate that are known to give error during loading. One benefit of the SmartChip system is the availability of a 36 markers × 144 samples format. The 36 markers × 144 samples format allows for 32 markers from the existing panel and also allows the addition of four markers for improving variety calls, while maintaining reasonable sample throughput.

Conclusions
In summary, we developed a high-throughput SNP-based protocol for the identification of Canadian wheat varieties to support quality assurance in the Canadian grain system. In this report, we show that updated markers (Perry and Lee 2015) provided improved accuracy of both the markers and the variety identification calls (Table S2). We also demonstrated that both the OpenArray and the SmartChip genotyping platforms can support this method. Both platforms have benefits and drawbacks in terms of cost, the length of processing time, throughput, accuracy, and available assay-by-sample formats. While not all of these variables are compared in this study, in general, we observed that the OpenArray system has higher throughput than the SmartChip system, but this will vary depending on the number of samples being tested. In contrast, the SmartChip system had improved accuracy and fewer "no calls" than the OpenArray system. The addition of four more assays to the previous 32-marker format of OpenArray is another benefit of the SmartChip system, particularly as related varieties are released that share similar marker patterns and there arises a need for additional markers. Nevertheless, both systems provide strength in identifying Canadian wheat varieties and offer unique advantages that may make them suitable to different users and applications.

Article information
History dates