Comparative analysis of ZoomAgri and DNA-based methods for the assessment of varietal purity in barley

Executive summary

This report presents the findings of a comparative analysis conducted by the Grain Research Laboratory to evaluate the accuracy and agreement of varietal purity assessments in barley samples using ZoomAgri’s image-based technology, relative to the current DNA-based genotyping standard method.

Varietal purity is a critical quality parameter in the barley supply chain, with malsters and buyers typically requiring a minimum of 95% purity to ensure product quality and avoid financial penalties. DNA-based methods are widely regarded as the gold standard for varietal assessment, due to their high specificity and broad varietal recognition. However, they remain costly and time-consuming. In contrast, ZoomAgri offers a novel approach, using artificial intelligence assisted image recognition, that significantly reduces both cost and turnaround time.

The Grain Research Laboratory analyzed 71 barley samples using both techniques. The samples represented 6 commonly grown Canadian malting barley varieties—AC Metcalfe, CDC Copeland, AAC Synergy, AAC Connect, CDC Fraser, and CDC Churchill—as well as one general-purpose variety, CDC Austenson.

Statistical analysis using Lin’s Concordance Correlation Coefficient (CCC) revealed strong agreement between the two methods, with CCC values reaching 0.98 and varietal purity estimates generally falling within a ±5% range. Bland-Altman plots further confirmed this agreement, showing mean differences close to zero and narrow limits of agreement, indicating minimal bias and acceptable variability. For specific varieties, such as AAC Connect and CDC Churchill, the image-based method by ZoomAgri showed a slight overestimate and underestimate of purity, respectively, but the differences in varietal purity estimates remained within a ±5% range.

Repeatability tests demonstrated high precision across replicates, reinforcing the reliability of ZoomAgri’s assessments.

These results indicate that ZoomAgri’s estimates of varietal purity for the 6 commonly grown Canadian malting barley varieties (AC Metcalfe, CDC Copeland, AAC Synergy, AAC Connect, CDC Fraser, and CDC Churchill) are both highly correlated and numerically consistent with the DNA-based results.

While ZoomAgri currently recognizes a smaller number of varieties compared to the DNA-based method, its speed, affordability and accuracy make it a promising tool for routine purity assessments, particularly in operational contexts where rapid decision-making is essential. Future validation across a broader range of barley genotypes and environmental conditions will further enhance its utility.

The results presented here reflect only the findings obtained using the specific version, database, and method available at the time of testing. The version, database, and analytical method used for image-based analysis may be subsequently updated by ZoomAgri. It should also be noted that the samples analyzed were received and tested across two distinct crop years.

Background

Varietal purity is a critical determinant of both quality and market value for malting barley. Domestic and international buyers typically require a minimum of 95% purity in purchase contracts. Failure to meet this threshold often results in financial penalties for growers and grain companies, highlighting the need for reliable, fast, and cost-effective variety identification methods.

Initial methods for barley variety identification focused on the extraction of proteins and their separation through techniques such as gel electrophoresis, high-performance liquid chromatography and mass spectrometry (Marchylo, 1987; Marchylo and Mellish, 1992). Despite their scientific merit, these methods lacked sufficient resolution due to the limited variation in grain proteins among elite barley varieties.

DNA-based methods offer significantly greater resolving power due to the abundance and specificity of DNA markers that can be developed for each barley variety. These markers are consistent across tissues and developmental stages of the grain and can be used to analyze malt—making them highly relevant to the brewing industry.

Today, DNA-based methods are considered the gold standard for varietal identification, valued for their accuracy, specificity, and reproducibility.

The Grain Research Laboratory has recently developed an enhanced DNA-based method using 24 custom TaqMan® genotyping assays analyzed via the SmartChip system (Takara Bio Inc.) (Sung-Jong et al., 2024). This method enables the identification of more than 124 Canadian barley varieties, maintains high accuracy, and offers flexibility, higher throughput, and better cost-efficiency, compared to previous methods. The major limitations are still relatively high cost and time of analysis.

• Cost: approximately $600 to $1200 per sample.
• Time: 24 to 48 hours per analysis.

ZoomAgri has introduced a novel approach to varietal identification using digital imaging and artificial intelligence to identify barley varieties. The major advantages are reduced cost and time of analysis.

• Cost: approximately $15 per sample.
• Time: approximately 3 minutes per analysis.

Recognizing its potential, the Grain Research Laboratory initiated an evaluation of ZoomAgri’s technology for its effectiveness and accuracy in assessing varietal purity in Canadian malting barley.

Objective

The objective of this study was to evaluate the agreement and accuracy of varietal purity assessments in barley samples using ZoomAgri technology, in comparison to the standard DNA-based method developed and currently employed by the Grain Research Laboratory.

Materials and methods

The Grain Research Laboratory obtained 71 barley samples, with 54 samples from the 2023 crop year and 17 samples from the 2024 crop year. These samples were acquired through the Canadian Grain Commission’s Harvest Sample Program, annual barley harvest survey, and cargo monitoring program. Table 1 provides the distribution of individual barley varieties included in this study.

Approximately 20 grams (equivalent to 250 to 400 kernels) were sub-sampled from each barley sample, placed on a gridded tray that kept kernels separate and scanned using the ZoomAgri instrument (Figure 1). Following analysis, all kernels were recovered and retained, and 108 individual recovered kernels were selected for DNA extraction and analysis.

DNA was extracted from individual kernels using a 96-well format, following the protocol described by Perry and Lee (2015). The resulting DNA samples, along with the TaqMan® genotyping assays, were loaded into the SmartChip system using the MultiSample NanoDispenser (Takara Bio Inc.) and combined with the SmartChip qPCR master mix (Takara Bio Inc.), in accordance with the manufacturer’s instructions and tested using the methods described in Sung-Jong et al., 2024.

A: Approximately 20 gram sample (250-400 kernels)

B: 108 kernels analyzed by DNA-based method

Results

Agreement between ZoomAgri and DNA-based methods in determining varietal purity of barley samples

The agreement between results obtained by ZoomAgri and DNA-based methods was assessed using the Lin’s Concordance Correlation Coefficient (CCC). Lin’s CCC is a statistical measure that is used to evaluate agreement between two sets of data—especially when comparing a new measurement method to a ‘gold standard’ or reference method. Lin’s CCC, which combines two key components: precision and accuracy, is designed to evaluate how well two sets of measurements agree—not just how strongly they correlate.

Table 3  Lin’s Concordance Correlation Coefficient values and corresponding levels of agreement^{Footnote a}

CCC value	Agreement level
> 0.99	excellent agreement
0.95 to 0.99	strong agreement
0.90 to 0.95	acceptable
< 0.90	weak or inconsistent
Table 3 Notes Table 3 Note 1 Based on guidelines proposed by McBride (2005). Return to table 3 note a

Figure 2 presents Lin’s CCC, illustrating the level of agreement between the percentage estimates of the major barley varieties obtained using ZoomAgri technology and those derived from DNA-based genotyping methods. The CCC value of 0.98, with a 95% confidence interval of 0.96 to 0.98, indicates substantial agreement between the two methods. This high concordance suggests that ZoomAgri not only correlates strongly with the DNA-based reference method but also produces numerically consistent results, supporting its reliability as an alternative tool for varietal purity assessment in the tested samples.

Figure 3 presents a Bland-Altman plot, another statistical method used to evaluate the agreement between two measurement techniques. In this plot, the x-axis represents the average percentage of each barley variety estimated by ZoomAgri and DNA-based methods, while the y-axis displays the difference between the two estimates (ZoomAgri% minus DNA%). The plot includes a mean difference line (referred to as bias), which indicates whether one method consistently overestimates or underestimates the results relative to the other.

The mean difference observed was close to zero, suggesting no systematic bias between ZoomAgri and DNA-based genotyping methods. The limits of agreement, defined as the mean plus or minus 1.96 standard deviations, ranged from –5% to +5%, indicating that differences in varietal purity estimates between the two methods were generally within a ±5% range. Notably, differences in estimating the percentage of minor varieties also fell within this range. This level of agreement supports the use of ZoomAgri as a reliable alternative for assessing barley varietal composition.

Figure 4 presents Bland-Altman plots for individual barley varieties, illustrating the agreement between varietal percentage estimates obtained using ZoomAgri technology and DNA-based genotyping. For AAC Connect, the mean difference line of +1.6% suggests that ZoomAgri tends to slightly overestimate the proportion of this variety in analyzed samples. Conversely, for CDC Churchill, the mean difference line of –1.3% indicates a slight underestimate by ZoomAgri. Across all varieties examined, the differences between the two methods were generally contained within a ±5% range.

Text description

Scatter plot of ZoomAgri results on the y-axis and DNA-based results on the x-axis for estimates of major barley varieties. The solid 45° line represents perfect agreement between the 2 methods, and the dashed line represents the observed agreement. The plot shows that there is a substantial agreement between the 2 methods.

Text description

Plot showing the difference between ZoomAgri and DNA-based results on the y-axis and the average result of the 2 methods on the x-axis for major and minor barley varieties. The mean difference line is approximately 0, indicating no systemic bias. The limits of agreement ranged from -5% to +5%.

AAC Connect

AAC Synergy

CDC Copeland

CDC Churchill

CDC Fraser

Text description

Plots showing the difference between ZoomAgri and DNA-based results on the y-axis and the average result of the 2 methods on the x-axis for individual barley varieties. The AAC Connect plot has a mean difference line at 1.6, indicating a slight overestimation by ZoomAgri. The CDC Churchill plot has a mean difference line at -1.3, indicating a slight underestimation by ZoomAgri. The mean difference line is at 0.1, -0.56 and 0.9 for the plots of AAC Synergy, CDC Copeland and CDC Fraser, respectively.

Conclusions

This study demonstrated that the image-based ZoomAgri technology provided a reliable and accurate assessment of varietal purity of 6 commonly grown malting barley varieties: AC Metcalfe, CDC Copeland, AAC Synergy, AAC Connect, CDC Fraser, and CDC Churchill. The high Lin’s Concordance Correlation Coefficient values, consistently above 0.95, indicated strong agreement between ZoomAgri technology and DNA-based genotyping in estimating the percentage of both major and minor barley varieties in analyzed samples. Additionally, Bland-Altman analyses revealed minimal bias and narrow limits of agreement, with differences generally within a ±5% range, further supporting the consistency of ZoomAgri’s performance.

While DNA-based methods remain the gold standard for varietal identification due to their broad recognition capacity and molecular precision, ZoomAgri offers a rapid and economical option for routine purity assessments. Its ability to deliver near-equivalent results with significantly reduced processing times makes it particularly valuable for quality control in the malting and brewing industries.

Future work may focus on expanding ZoomAgri’s varietal recognition library and validating its performance across a wider range of barley genotypes and environmental conditions. Nonetheless, the current findings underscore its potential to complement traditional genotyping methods in specific operational contexts.

References

Marchylo, B.A. 1987. Barley cultivar identification by SDS gradient page analysis of hordein Canadian Journal of Plant Science 67: 927.

Marchylo, B.A. and Mellish, V.J. 1992. The development and application of varietal identification technology at the Grain Research Laboratory. Grain Research Laboratory 65th Annual Report.

McBride, G. B. 2005. Using Statistical Methods for Water Quality Management: Issues, Problems, and Solutions. New York: John Wiley & Sons.

Perry, D. and Sung-Jong L., 2015. Identification of Canadian wheat varieties using OpenArray genotyping technology. Journal of Cereal Science. 65: 267.

Sung-Jong L., Eckhard, M., Dusabenyagasani, M., Izydorczyk, M., Demeke T., Perry, D., and Walkowiak, S. 2024. Identification of Canadian barley varieties by high-throughput SNP genotyping. Canadian Journal of Plant Science. 104: 388.