Bioflow gets smarter with new modules and cloud features

The development of Bioflow, CGIAR’s open-source breeding analytics pipeline, funded by GIZ through Crops to End Hunger, has never slowed down since its launch last year. Following several hackathons in February 2024 and November 2024, staff from all CGIAR Centers contributed their expertise to the ongoing development of the tool, highlighting the collaborative DNA that underpins its development. 

Here are three major upgrades currently in development – which are expected to increase the power and ease of use of Bioflow for CGIAR and partner crop breeders.

Better genotype × environment modelling with ASReml integration 

Angela Pacheco, Associate Scientist, Applied Agricultural Statistics/Biometrics at CIMMYT, is currently leading the integration into Bioflow of ASReml, a licensed R package recognized as the gold standard for modelling multi-environment trial data. This integration responds to a long-standing request from breeders: more precise and flexible modelling of genotype × environment (G×E) interactions. 

“ASReml allows breeders to move beyond the earlier assumption in Bioflow that genetic covariance between pairs of environments is uniform; a simplification often at odds with reality. With this new module, Bioflow will be able to more accurately reproduce the complexity of real-world growing conditions, making selection models more relevant and robust”, Angela said. 

For example, models with different variance structures such as unstructured, diagonal, and factor analytic can now be incorporated, and, at the same time, the genetic relationship matrix with markers and pedigree could be added. 

ASReml is licensed software, and its high performance and wide acceptance among breeders make it a valuable addition to Bioflow. Integration is already in development and will soon be available on the production server.

Improving genomic selection in clonally propagated crops

Another update under way is genomic prediction of cross-performance, a method that changes the way breeders select parents in clonal crops like cassava, potato and sweet potato. 

Lorena Batista, Quantitative Genetics Specialist is building on previous research led by Christian Werner, to develop a module that helps breeders select the best crosses for increasing rates of genetic gain, rather than truncation selection of individual parents. This is important because in clonal crops, non-additive effects can play an important role in trait expression, but they are not directly considered when selecting parents based on genomic estimated breeding values (GEBV). 

“Genomic Prediction of Cross Performance ensures that both additive and dominance effects are taken into account when selecting crosses. This means not only that we increase the frequency of favourable alleles, but also that we preserve the best allele combinations within the breeding population. This is particularly important for clonal crops, where non-additive variation is usually high and inbreeding depression can seriously impact trait performance”, Lorena explained. 

Using cross-performance prediction promotes long-term genetic gain while minimizing the buildup of inbreeding in breeding populations over time. Several roots, tubers, and bananas (RTB) programs are already exploring how to apply this approach and incorporating it into Bioflow makes it even more accessible to RTB breeders striving to accelerate genetic improvement.  

Cloud storage to enhance usability and access

Bioflow is going all cloud-native, not just for processing, but now also for storage. Khaled Al-Sham’aa, Associate Scientist in Biometrics at ICARDA, in collaboration with the Enterprise Breeding System (EBS) team, is finalizing a new cloud storage feature. Until now, Bioflow users had to save analysis outputs on their local machines, which could be slow due to bandwidth constraints and made it harder to resume or share analyses later, creating barriers to seamless work across sessions or devices. 

With this new functionality, every user on the production server will have access to personal cloud storage. Analyses will be saved securely and can be accessed anytime, from anywhere, streamlining workflows and making data-driven decision making much more user-friendly. 

“Bioflow’s calculations already run on Amazon Web Services (AWS) servers. By moving storage to the cloud, even users with limited computing power will be able to perform all their tasks online, from running analyses to storing results. This is precisely the main goal of Bioflow: to fill gaps in computing capacity and standardize research quality”, Khaled explained. 

To date, 73 national breeding programs regularly use Bioflow. 

These updates reflect the growing momentum behind Bioflow as a global public good. By integrating new, recognized tools and reducing technical barriers, the platform keeps empowering CGIAR and partner breeders worldwide with smarter, faster, and more scalable breeding analytics.  

Stay tuned for more updates as these features are rolled out to the production environment!

Resources

• More about Bioflow, CGIAR Breeding Analytics Pipeline 

• Werner CR, Gaynor RC, Sargent DJ, Lillo A, Gorjanc G, Hickey JM. Genomic selection strategies for clonally propagated crops. Theor Appl Genet. 2023 Mar 23;136(4):74. doi: 10.1007/s00122-023-04300-6. PMID: 36952013; PMCID: PMC10036424.

***

Writen by Julie Puech for Breeding for Tomorrow. Main image: CGIAR Breeding Analytics Pipeline hackathon, Nairobi, 2025. Credit: Breeding Resources. Crops to End Hunger is funded by the German Federal Ministry for Economic Cooperation and Development (BMZ) commissioned by the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) through the Fund International Agricultural Research (FIA).