GENESIS: HPC-Enhanced Genomic Selection for Climate-Resilient CropBreeding

Crop breeding programmes typically operate ondevelopment cycles of approximately seven years,from earlyselection to variety release. Underaccelerating climate change, this creates a structuralmismatch: new varieties entering the market arelargely optimised for historical climatic conditionsobserved at the start of the breeding cycle, rather thanfor theenvironmental conditions they will face duringcultivation. As climate variability increases, thistemporal gap reduces the reliability of traditionalmulti-environment trials and limits breeders’ ability,particularly SMEs, to robustly evaluate genotypeperformance under future stress scenarios.

The GENESIS Business experiment, led by bSpokeSolutions, addresses this challenge by developing anHPC-powered genomic-selection platform designed toaccelerate and modernise the evaluation and selectionof breeding material for climate-resilient crops. Byintegrating large-scale genomic, phenotypic,environmental, and multiomics-ready datasets withadvanced machine-learning and deep-learningapproaches, GENESIS enables scalable, data-drivenevaluation, ranking,and selection of breeding lines,supporting the prediction of genotype performanceacross heat-and water-stressed environments.Leveraging EuroHPC systems, bSpoke processes large-scale datasets, trains and optimises ML/DL models,and validates them usingreal-world seed-improvement data from BIOS Agrosystems.The GENESIS Platform, will be delivered as Software-as-a-Service (SaaS), enabling agricultural SMEs andresearch institutions to access high-precision breedinganalytics without requiring in-house HPC infrastructure.

Presentations

Organisations involved

Short description of the experiment

GENESIS establishes an HPC-powered experimental pipeline supporting advanced breedinganalytics for climate-resilient crops. Led by bSpoke, the experiment integrates large-scalegenomic, phenotypic, environmental, and multiomics-ready datasets with regional climateinformation, including CORDEX-based projections. Using HPC-scale CPU and GPU resources,GENESIS develops and trains machine-learning and deep-learning models focused onlearning genotype–environment relationships and predicting genotype performance acrossdiverse and climate-stressed environmentsthereby supporting the development of moreclimate-resilient crop varieties.The GENESIS experiment includes large-scale data ingestion, preprocessing, exploratoryanalysis, model development, and GPU-acceleratedmodel training executed on EuroHPCsystems, followed by validation using real-world multi-environment field-trial data fromBIOS Agrosystems (BIOS). The resulting models are deployed through a user-friendlyplatform delivered as Software-as-a-Service (SaaS), enabling both turnkey breeding analyticsand configurable evaluation workflows. The GENESIS experiment demonstrates how HPCand AI can significantly accelerate selection and evaluation phases in breeding programmes,while establishing a robust foundation for the progressive integration of additional genomicand multiomics predictors, paving a scalable digital pathway toward next-generation,climate-smart crop improvement.

Outlook

Building on the outcomes of the experiment, future developments will include:

• Expansion of GENESIS to additional crops and broader multi-environment breedingtrials.

• Incorporation of temporal climate scenarios to support longer-term genotypeperformance forecasting under evolving climate conditions.

• Continuous optimisationof ML/DL pipelines using EuroHPC resources.

• Establishment of a commercial subscription service targeting SMEs lacking access toadvanced computing infrastructure.
• Exploration of collaborative partnerships for international deployment

Lessons learned

• HPCis essential for handling high-dimensional genomic and multiomics datasets,enabling scalable preprocessing, feature engineering, and model optimisationworkflows.
• GPU acceleration reduced deep-learning training times from weeks to hours, enablingrapid iteration.
• Parallel processing allowed scalable integration of climate data, genomic information,and phenotypic observations.
• Efficient storage and data-management strategies were critical for handling largevolumes of raw and processed datasets.
• HPC enabled systematic hyperparameter optimisation would be infeasible on localinfrastructure.
• First-time HPC users, bSpoke and Bios, were able to adopt HPC workflows efficientlywith support from EuroCC@Greece.

Expected impact

Society

GENESIS contributes directly to food-system resilience by enabling rapid development ofcrop varieties that withstand climate extremes. Breeding cycles can be shortened from 7-10years to 3-5 years, allowing farmers to adapt more quickly to changing conditions. Theresulting resilient varieties reduce crop losses, stabilize food supply chains, and supportsustainable farming with lower chemical inputs.

Business

For BIOS and other SMEs operating in the seed sector, GENESIS delivers direct operationaland strategic benefits by reducing the effort and time required for early-stage selection andmulti-environment evaluation. Through AI-driven analytics and access to HPC-enabledmodelling via a cloud-based platform, SMEs can improve the quality and robustness ofselection decisions without proportionally increasing field trial costs. GENESIS enables moresystematic assessment of genotype performance under climate variability, lowering barriersto advanced modelling and strengthening the competitiveness of SMEs against larger seedcompanies with greater internal resources.For bSpoke, GENESIS represents a scalable commercial offering that translates advancedHPC and AI capabilities into a Software-as-a-Service platform for digital breeding analytics.The platform creates recurring revenueopportunities through subscription-based access,while positioning bSpoke as a technology provider operating at the intersection of high-performance computing, artificial intelligence, and digital agriculture.

Exploitation roadmap

The exploitation of GENESIS follows a structured roadmap centred on platformcommercialisation and strategic collaboration. bSpoke will lead the exploitation of GENESISthrough its deployment as a cloud-based Software-as-a-Service (SaaS) platform providingAI-enabled breeding analytics and support for genomic-selection workflows. The platformtargets BIOS and other SMEs operating in the seed sector, as well as research organisationsand agricultural innovation actors that lack in-house access to advanced computing andmodelling capabilities. In parallel, bSpoke will explore strategic collaborations with Agritechand Biotech stakeholders to support crop-specific extensions, regional deployment, and co-funded research and development activities. These collaborations may include jointdevelopment or revenue-sharing models tailored to specific crops or markets, whileleveraging the scalable and modular architecture of the GENESIS platform.