10 March 2026, AU: New digital tools are set to unlock the potential of plant genetic resources stored in the Australian Grains Genebank to accelerate the development of improved varieties.
Rugged up against the minus 20° C temperature inside the Australian Grains Genebank (AGG) at the Horsham SmartFarm, Dr Sally Norton is dwarfed by floor-to-ceiling shelves filled with tiny boxes.
In them are some 217,000 accessions – collections of seeds that represent grain crops from around the world, including cereals, pulses, oil seeds and legumes.
As a joint investment between the Victorian Government and GRDC, the AGG is home to one of the largest global collections of plant germplasm.
As an active resource, it distributes more than 30,000 samples each year to pre-breeding research and commercial breeding programs to aid the development of crops with greater resilience and productivity.
And now, new digital tools are lifting the potential of the genebank’s genetic resources to more quickly deliver improved varieties.
“The suite of digital tools that has been developed will help researchers and breeders identify material that has high-value characteristics relevant to their objectives, ranging from climate resilience to disease resistance and for functional food such as high-protein pulses,” says Dr Norton, who is the AGG Research Leader and Curator.
World of possibility
The advances are part of the 5-year $30 million AGG Strategic Partnership between GRDC and Agriculture Victoria, established in 2022 with a mission to unlock the genebank’s genetic potential.
One of the first steps in the investment is the world’s largest genebank DNA genotyping exercise.
This project, led by Gabriel Keeble-Gagnère (the AGG Science Program Leader), is characterising the genomes of about 75% of the AGG’s target species, including all major cereal, pulse and brassica crops. In total, this represents more than 100,000 accessions, with genotype data for roughly half already released for public access.
Genotyping the collection is a significant advance on the existing system, which attaches ‘passport’ information to seed samples. While passport data includes crop variety and origin, information on characteristics such as disease resistance can be limited, Dr Norton says.
Once paired with genomic data, a “world” of genetic possibility opens, she says. “A curse of genebanks everywhere is that this passport information can be lacking.
But when samples are genotyped, you can screen them for whatever particular trait you’re interested in and understand what it may or may not contain.
For example, a genebank might house germplasm with traits that could improve acid soil tolerance in lentils, she says. “But this may never be identified unless there is a way of finding it.”
This is where the new suite of digital tools will play a critical role.
Faster selection of the best material
The world-leading tools have been developed, in collaboration with industry, by Mr Keeble-Gagnère’s team at Agriculture Victoria Research. They can be used to explore genetic resources stored in the AGG as well as other genebanks globally to pinpoint germplasm of interest.
“They enable a thorough desktop evaluation of the genetic material we hold to better understand what it might be useful for,” Dr Norton says. “And then it can be tested in research and breeding pipelines to see if it is as good as predicted.”
Mostly named after baked goods, the tools are designed to complement and interact with each other (see Figure 1).
Digital tools
Among them are Brioche, Shortbread and the more literally named Genolink – a ‘middleware’ data management tool that seamlessly connects passport data accessed through an online database of global genebank accessions to genotype databases.
In a few clicks, researchers and breeders can select accessions based on passport data and immediately query the genotypic data or export for further analysis.
This helps industry and also genebank managers to identify, align and update genetic material in their collections, ultimately improving access and functionality for end-users, says Dr Norton, who uses the tool daily.
“For example, a researcher or breeder might be seeking chickpea material with Ascochyta resistance,” she says.
“Out of 10,000 chickpea lines, we might have 500 that carry Ascochyta resistance based on passport information – and Genolink can help identify them and extract their genotypic data. So, it’s a much more targeted and strategic use of material.”
Similarly, Brioche and Shortbread are transformational tools that reformat and organise genetic and genomic data to make them interoperable with existing datasets.
Pinpointing traits
This data can then be fed into the custom designed tools such as Pretzel, Lamington and Fairy Bread, which provide further insight into germplasm with particular traits.
Pretzel, for example, allows genotype data to be used to identify accessions with traits of interest. This means that if a breeder is seeking yellow rust resistance in bread wheat, Pretzel allows you to identify resistant accessions from known trait-linked markers.
It also identifies accessions with potentially novel sources of resistance based on unexpected genotype patterns in regions known to be associated with resistance.
“Pretzel unlocks a whole new capability, and there’s nothing else that exists that does what this suite of tools can do,” Dr Norton says.
In other developments, the Lamington tool can identify subsets (from a much larger number of accessions) that capture genetic diversity.
“So you could ask it for the best 200 chickpea lines with Ascochyta resistance, and it will create a subset that is more usable in research or breeding programs,” she says.
“You can’t put 20,000 lines in the field and easily evaluate them, but you could much more easily manage 200.”
These subsets or other genetic material samples can be further explored with the Fairy Bread tool. Fairy Bread combines passport and genotype data, enabling the genetic relatedness of genebank accessions to be visualised at the same time their passport data is explored.
“You can drill down to understand more about the ‘why’ behind a particular trait,” says Dr Norton.
“So if you were looking to develop lentils for acid soils, you could potentially target genetic material that has evolved under those pressures.”
Plain answers
Overarching all the tools is Bakery – an artificial intelligence module currently under development that leverages large language models and knowledge graphs to enable the interrogation of data with plain language queries.
Bakery has the potential to transform how genebank users access information, allowing them to focus on their particular question while letting Bakery work out the best way to answer it.
Pre-breeding researchers and commercial entities are already using the data released by the Partnership, which is free with attribution under a Creative Commons licence. There is also a range of documentation and training programs available on how to use the new tools.
Dr Norton says she is excited about the technology and how it may change the way genebanks are utilised now and into the future.
“There’s a whole lot of material in genebanks and these tools enable questions to be asked of that material. It will be interesting to see what genetic material pops up with high value for certain traits – particularly those that have never before been distributed.”
Global Agriculture is an independent international media platform covering agri-business, policy, technology, and sustainability. For editorial collaborations, thought leadership, and strategic communications, write to pr@global-agriculture.com
