Calculating Accuracy and Reliability of Random Effect Estimates with ASReml-R

Enhancing agronomic research through precise statistical analysis with Genstat

Deben Agronomy is dedicated to giving potato and vegetable growers the right guidance to grow better crops. They use their expert knowledge gained through independently generated trial data to provide actionable insights that boost crop yields and promote sustainable practices.  ### The challenge: Cultivating confidence In the pursuit of their research endeavours, Deben Agronomy faced a pressing challenge. They needed robust statistical analysis for their vegetable field trials data, specifically focusing on techniques like analysis of variance. Their primary goal was to instil confidence in the results derived from experimental trials—a critical factor for their decision-making and for their clients too.   For a field research scientist like those at Deben Agronomy, having confidence in their findings hinges on the ability to definitively attribute differences in results to specific treatments rather than mere chance. Achieving this level of accuracy was paramount in maintaining the integrity of their research and recommendations. To meet this challenge, they turned to Genstat.  ### Choosing the best tool: The decision process Finding the right solution involved a meticulous evaluation of available options. While considering various alternatives, Deben Agronomy's experts harkened back to their prior experience with Genstat. The intuitive nature and ease of use they had encountered in the past left a lasting impression. As a result, despite exploring other choices, Genstat's familiarity and reliability made it the clear choice.  ### Hello Genstat: Where stats become user friendly While acknowledging that they only scratched the surface of Genstat's capabilities, Deben Agronomy's scientists praised the software for its swiftness in generating intricate statistical analyses. This efficiency not only bolstered their results but also facilitated clearer explanations for their clients. The ability to present data-driven insights with clarity became an asset that set them apart.  _“Ease of use, combined with excellent support, has made our experience with Genstat truly satisfying. It generates statistical analyses quickly and easily to support my results, and the graphing capabilities allow me to explain these results clearly to clients.”_ **Angela Huckle, lead R & D agronomist at Deben Agronomy** ### Harvesting the results: Improved reputation and trust Using Genstat helped to transform Deben Agronomy’s research outputs. They gained a higher level of credibility and could confidently attribute variations to specific treatments where there was a response from either the crop or pest. This precision enhanced their reputation and fostered greater trust among clients who depend on their guidance.   Deben Agronomy's partnership with VSNi's Genstat has shown how reliable statistical analysis tools can reshape the landscape of agronomic research. They fortified their methodologies and elevated the calibre of insights provided, contributing to the growth and sustainability of the agricultural industry as a whole.  <table style="border-color:hsl(240, 75%, 60%);border-style:double;"><tbody><tr><td><strong>Ready to enhance your agronomic research with Genstat or have questions about our services?&nbsp;</strong><br><strong>Our experts are here to assist you. Feel free to contact us&nbsp;</strong> <a href="https://vsni.co.uk/contact-us">https://vsni.co.uk/contact-us</a></td></tr></tbody></table>

Uncovering disease control strategies: A Genstat case study on Maedi-Visna (MV) epidemiology in sheep

This case study focuses on the collaborative research of [Andrew Illius](https://www.ed.ac.uk/profile/andrew-illius), Emeritus Professor of Animal Ecology at the University of Edinburgh, and [Nick Savill](https://www.maths.ed.ac.uk/school-of-mathematics/events/math-biology/mathbiol/nick-savill), Senior Lecturer in the School of Biological Sciences at the University of Edinburgh. They have been investigating the epidemiology and control of maedi-visna (MV), a debilitating lentivirus disease that infects sheep, goats, and wild ruminants causing inflammatory lesions to build up in the lungs, mammary gland, brain and the synovial membrane of joints. MV is a significant concern for farmers due to its rapid and silent spread within flocks, leading to welfare and productivity issues. By the time clinical signs are evident, the disease is well-established, with perhaps 30-70% of the flock infected. Through the use of Genstat, Andrew and Nick analysed long-term unbalanced experimental data, revealing crucial insights into disease control strategies. ### Research objective For a farmer, the disease raises crucial questions: How quickly will symptoms develop, and will welfare and productivity suffer? How soon will it spread through the rest of the flock? Is living with the disease more expensive than trying to eradicate it? The primary aim of this research was to find ways to control MV by analysing data from long-term experiments. MV is caused by a small ruminant lentivirus (SRLV). However, despite a substantial amount of research on the pathology, immunology, and molecular biology of SRLVs, the team were surprised to find there was a lack of quantitative analysis on SRLV epidemiology. ### Modelling MV transmission The research team leveraged a mathematical model developed by Nick Savill based on infectious disease epidemiology, and the known and unknown aspects of SRLV biology. This model estimated the probability of seroconversion, shedding light on the disease's transmission patterns. The study was the first to estimate transmission parameters, revealing that MV is strongly associated with housing, and transmission is negligible among sheep kept at grass. This finding presented an opportunity for disease control exploiting the fact that transmission of the virus is too slow between grazing sheep to sustain the disease. ### Understanding disease spread Further analysis focused on understanding infectiousness changes during an individual's infection. The team examined whether ewes with long-standing infections posed the greatest risk to uninfected sheep, and also explored management strategies to slow disease spread. Genstat was instrumental to this analysis. ### Quantifying costs with Genstat To assess the costs of living with MV, a large dataset from [Lublin University of Life Sciences](https://up.lublin.pl/en/university/) was analysed. This dataset included lambing records from about 800 ewes of different breeds, with over 300 ewes present each year. Given the complex nature of the data (unbalanced design, multiple observations from individual ewes and rams over several years, and different breeds) the team utilised Genstat to analyse it. The team employed mixed models to distinguish random and fixed effects, making use of the latest release of Genstat, which included Akaike’s Information Coefficient (AIC) for model selection. The analysis addressed pre-weaning lamb mortality and showed that the probability of mortality was strongly affected by lamb birthweight and the SRLV status of the ewe. Lamb growth rate was also shown to be negatively related to the time since the dam had become infected. ### Genstat's role in the study Andrew Illius, with decades of experience using Genstat, praised the software for its crucial role in agricultural and ecological research.  “_The software's evolution from a Fortran-based system to a menu-driven environment on Windows significantly sped up analysis processes. Genstat's ability to handle unbalanced data and mixed models proved essential in this study. It’s a fantastic product, and a pleasure to use_.” ### Advancing scientific knowledge through Genstat The research provides evidence-based insights for controlling MV, suggesting targeted disease control measures and social distancing strategies for sheep. This Genstat case study demonstrates the software's versatility in analysing complex data and its importance in advancing scientific knowledge. By utilising Genstat, the collaborative research of Andrew Illius and Nick Savill contributes to understanding Maedi-Visna epidemiology, potentially leading to more effective disease control measures for the benefit of farmers and livestock alike. --- ### Example output from Genstat 

Accelerating sugar beet breeding insights for SESVanderHave

### Analysing large phenotypic and genotypic datasets using the H3-pipeline Belgium-based [SESVanderHave](https://www.sesvanderhave.com/?redirect=yes) stands tall as a global leader in sugar beet breeding, selling billions of seeds annually. With their commitment to innovation and quality, they have successfully developed localised sugar beet varieties for individual markets in over 50 countries. The company uses a variety of cutting-edge technologies, including the [H3-pipeline](https://vsni.co.uk/services/h3-analytics), to develop new varieties of sugar beet that have higher yields and improved resistance to pests and diseases. ### Managing the challenge of big data in plant breeding SESVanderHave faced significant challenges due to the exponential growth of data in plant breeding. With new candidate varieties being tested each season, characterised in greater detail and incorporating phenotypic and molecular marker data, they needed a powerful solution that could model and condense this vast amount of information for effective decision making. ### A new statistical tool for sugar beet breeding Recognising these challenges, the [H3 consortium](https://h3.ac.uk/), comprising VSNi, [Biometris](https://www.wur.nl/en/Research-Results/Research-Institutes/plant-research/biometris.htm) and [JHI](https://www.hutton.ac.uk/), collaborated with SESVanderHave to develop a cutting-edge software pipeline. This innovative pipeline seamlessly integrates data and models, enabling breeders to obtain relevant information essential for their work. The pipeline encompasses various features, including the design and analysis of phenotyping experiments, allele transmission from parents to offspring, coupling of phenotypes to genotypes in QTL mapping and GWAS models, and genomic prediction models. One of the key requirements from SESvanderHave was the ability to analyse vast phenotypic and genotypic datasets swiftly. The H3-pipeline needed to exhibit exceptional speed and robustness: the DNA of thousands of young plants are genotyped, and based on advanced genomic prediction methods the best new genotypes are selected and field tested in numerous global locations to ensure their performance under diverse conditions. ### The H3-pipeline: essential components and processes The H3-pipeline is a crucial component of the breeding program at SESVanderHave, comprising different modules that cover key steps in the selection of lines. The process begins with examining the trial field layout and generating an optimal experimental design using R. Next, the pipeline analyses field trials using the SpATS R package to calculate Best Linear Unbiased Estimates (BLUEs) and heritabilities and identify potential outliers. Subsequently it creates detailed reports for each trial, summarising the results for all genotypes and their important traits. Next it computes Best Linear Unbiased Predictions (BLUPs) across environments using ASReml-R, while employing genomic selection to identify new genotypes based on a comprehensive training set. Additionally, the pipeline incorporates a crossing tool that uses genomic predictions to identify the ideal parent combinations for the upcoming breeding cycle. The breeding pipeline's analytics were developed by Biometris in the R programming language using the ASReml library. To ensure flexibility and modularity, VSNi implemented the pipeline using a microservices architecture. This approach allows independent development and deployment of different modules within the pipeline. For efficient management and execution of the pipeline's modules, JHI developed a dashboard using a combination of Java and Vue.js. The dashboard is designed to run seamlessly in Docker environments, allowing it to be easily scaled up as required.  _The dashboard of the SESVanderHave H3 Analysis Pipeline_ ### Higher yields, improved disease resistance Agriculture is currently undergoing rapid and profound changes, such as the consumer-driven reduction in demand for plant protection products and fertilisers. This trend is reinforced by climate change, which has already led to an expansion of the occurrence of plant pathogens. In addition, climate change causes more extreme weather conditions, such as extended heat and drought events with significant impacts on crop yields. This plethora of changes to the agricultural context confronts breeding companies like SESVanderHave with an abrupt market-demand shift towards novel crop varieties with a higher level of tolerance and/or resistance against a wide range of pathogens as well as environmental stresses like drought and heat. At the same time, yield and agricultural output must remain at a competitive level to maintain profitability across the value chain. SESVanderHave’s vision for the development of modern sugar beet varieties is to enable the prediction of variety performance in both known and unknown environmental conditions. This predictive ability relies on a combination of genotypic data, environmental data and a priori physiological knowledge derived from yield trait components, most of which are obtained using new image capturing technologies. Over the past ten years, SESVanderHave has intensified its focus on data-driven and predictive breeding, integrating different types of data in the selection process. This has been made possible due to the reduced cost per molecular data point, advancements in phenomics, and the availability of real-time weather data. As a result, we have been able to implement a predictive breeding strategy that allows us to estimate the future performance of our varieties in different target environments. Hendrik Tschoep, Director of Breeding at SESVanderHave explains their direction:  “_It was clear to us that the main challenge was not generating new data but being able to extract the meaningful information needed to maximise the genetic gain in our breeding programs. Therefore, when we began our collaboration with the H3 consortium one of the main challenges was to develop a robust, user friendly, semi-automated statistical pipeline that would allow our breeders to integrate the most relevant data to select the optimal varieties of the future. The H3 pipeline makes it extremely simple, allowing us to utilise different types of data across years, locations, and experiments to select the truly superior offspring in our breeding nurseries.”_ The H3 pipeline has helped SESVanderHave to develop new varieties of sugar beet that have higher yields and improved resistance to disease and pests. As a result, SESVanderHave has been able to accelerate the genetic gain for sugar yield coupled with the introgression of new disease traits at the same time, responding to market demands and improving profitability. ### Shaping the future SESVanderHave's collaboration with H3 Consortium has paved the way for groundbreaking advancements in sugar beet breeding. The implementation of the H3 pipeline has empowered SESVanderHave to overcome challenges, develop superior varieties, and shape the future of sugar beet breeding. _“At SESVanderHave we are committed to providing high-quality, sustainable and profitable varieties to our customers. The H3 pipeline does the heavy analytical lifting for us so we can focus on the correct interpretation of the data and selection of the varieties of the future.”_ **Hendrik Tschoep, Director of Breeding, SESVanderhave** Juan Vegas, senior breeder at SESVanderHave added,  “_I’m a sugar beet breeder with more than 10 years of experience at SESVanderHave, and thanks to the new statistical H3 pipeline I’m able to combine different types of data easily, run my own predictions and even simulate new crosses without having to ask our data scientists. It just gives me a lot of freedom and new insights within my breeding programs, helping to unravel the true potential of our genetics._    _After years of close collaboration with the H3 consortium, we were able to solve the scientific challenges in breeding data analysis thanks to the long-standing experience of our partners in this field. Most importantly, H3 was able to translate the scientific findings into an analysis pipeline that is stable, reliable and requires minimum intervention, serving a commercial breeding program which has different requirements than tools used in academic science.”_  <table style="border-color:hsl(240, 75%, 60%);border-style:double;"><tbody><tr><td><strong>Would you like to explore how our consultancy services can drive results for your breeding programme? For more information or to dive deeper into the power of H3 Analytics, visit</strong> <a href="https://vsni.co.uk/services/h3-analytics">https://vsni.co.uk/services/h3-analytics</a>.</td></tr></tbody></table>