Achieve Your Breeding Goals With ⨉SeedScore.
Dramatically improve Breeding Efficiency and Genetic Gain!
View on YouTube
At Computomics, we understand the challenges encountered by plant breeders. New crop varieties have to be produced faster, cost-efficiently, and under rapidly changing climate conditions. In response to these challenges, our cutting-edge machine learning-based technology, ⨉SeedScore®, is here to empower you like never before. It's an invaluable addition to your plant breeder's toolbox.
Our tailored technical support and consulting services are designed to cater to your unique plant breeding program's needs, putting you on the path to a new commercial product. Let us support you in making more (and better) data-driven decisions on the path to creating your future signature genetics.
Benefits
⨉SeedScore® transforms siloed breeding data into actionable insights. By integrating genotypic, phenotypic, and environmental streams, our machine learning pipeline optimizes your predictive accuracy:
While the ⨉SeedScore® engine automates the complexity of machine learning and genomic prediction, our interactive dashboard translates these sophisticated outputs into an intuitive visual workspace.
Our machine learning-based technology ⨉SeedScore® is well-suited for a wide range of crops - field and vegetables crops for variety development as well as for hybrid production. We invite you to explore its applicability within your breeding cycle and discover the ways in which it can support your selection processes.
At a strategic level, variety development breeding cycles typically encompass three fundamental phases - the creation of genetic variation, the forming of varieties, and the testing - to identify the top performers resulting in a new commercial variety.
With our machine-learning-based technology ⨉SeedScore®, we can provide invaluable support in the early stages. Here, we excel at identifying e.g. the crosses that show the greatest potential to outperform existing commercial varieties. Based on phenotypic and genotypic data from your current breeding program, we predict the performance of the genetics available to you in new crosses. This foresight allows you to focus your efforts on the crosses and varieties with the highest probability of success.
As you approach the end of the variety forming phase, just prior to embarking on the testing phase, our expertise comes into play again. By using genetic information about your potential new varieties, we can predict their performance under field conditions. This predictive capability allows you to rationalize your field resources and focus on a selected subset of potential new varieties, greatly improving your chances of success in the final testing phase.
At a strategic level, hybrid breeding cycles typically encompass three fundamental phases - the creation of genetic variation, the creation of parental lines, and the creation and testing of hybrids - identifying the top performers resulting in a new commercial hybrid.
In all instances, a comprehensive understanding of the genetics of the hybrid parents and the phenotypic attributes of the hybrid is imperative.
With our machine learning-based technology ⨉SeedScore®, we can provide invaluable support in the early stages. Here, we excel at identifying the crosses that show the greatest potential to become parental lines. Based on phenotypic and genotypic data from your current breeding program, we predict the performance of the genetics available to you in new crosses. This foresight allows you to focus your efforts on the crosses and parents with the highest probability of success.
As you approach the end of the parental line forming phase, just prior to embarking on the testing phase, our expertise comes into play again. By using genetic information about your potential hybrid parents, we can predict their performance in all hybrid combinations under field conditions. This predictive capability allows you to rationalize your field resources and focus on a selected subset of potential new hybrids, greatly improving your chances of success in the final testing phase.
⨉SeedScore® predicts trait performance for material and environments you have not tested, helping you prioritize lines and crosses, decide where to test, and focus resources on the most promising candidates. Field observations remain essential: ⨉SeedScore® is designed to make testing more targeted, not to replace biological validation.
Yes. The ⨉SeedScore® modeling framework can be applied across crops and breeding systems, but the models themselves are not generic. We adapt the modeling strategy and validation approach strictly to your crop, target traits, and prediction question, making it suitable for any breeding strategy and production environment.
No. Our ML experts build and manage the models and computing infrastructure, while our breeding-informatics analysts work closely with your team to translate your goals into the right analyses. You receive ready-to-use results through intuitive tables and interactive visualizations without having to build or operate machine learning infrastructure yourself.
To get started, we typically need genomic marker data, measured target traits, and reliable identifiers linking the two. Environmental, weather, soil, or management data can be added when they help answer a specific breeding question, such as predicting genotype-by-environment effects, stability, new locations, or future conditions. This environmental data can be provided directly from your in-house weather stations or sourced from publicly available datasets.
We validate ⨉SeedScore® models against data they have not seen during training and choose the appropriate validation approach to reflect your real breeding decisions. Performance is assessed using tailored metrics for the specific trait and prediction question, so you can see exactly how reliably the model performs before using its predictions for selection.
⨉SeedScore® complements your existing breeding workflow and field trials. It operates as a recurring assess, train, predict, and select cycle, where incorporating new seasons of data continuously improves the model's accuracy over time. To provide full transparency, every intermediate step is visualized in an interactive dashboard. Your team can explore quality control, training and prediction results, internal and real field validations, as well as additional analyses. Ultimately, the dashboard empowers you to easily compare predictions, rank candidates, and export your final selections.
Your data is transferred securely and stored on Computomics-managed infrastructure in Germany (EU). Customer data is kept separate and is never shared between customers. You retain full ownership of your data and the resulting predictions, while the underlying ⨉SeedScore® algorithms and software remain Computomics' intellectual property.
Our unique climate-smart breeding technology ⨉SeedScore® supports breeders in answering some of global agriculture’s most pressing problems – increasing the diversity, quality, resilience and adaptability of future crops. It enables breeders to develop crop varieties for specific climates and changing environments by providing AI-based breeding decision support. Applying ⨉SeedScore® to your breeding program and leveraging all available data means dramatically improvign efficiency in every step you take and every decision you make.
