Applications

Six applications breeders are using today.

Genomic trait prediction is not one application — it depends on your crop, your trait target, and where in the selection cycle you're working. Below are the six use cases we see most often, with the data inputs, method, and expected output for each.

Use case 01

Drought tolerance screening before field entry

Maize Wheat Sorghum

Identifying drought-tolerant candidates before costly field trials in water-stressed environments requires multiple seasons of observations. Our Drought Tolerance Index score is derived from sequence embeddings validated against phenotype records from CIMMYT, GRIN, and internal program data. Submit VCF or FASTA — receive a ranked candidate list within hours.

Typical use: reduce a 800-line candidate pool to 40–50 entries for first field entry season. Selection is based on ranked Drought Tolerance Index scores — not gut instinct, not a single marker, not a preliminary trial. The shortlist goes into the field with a predicted score and a confidence interval already attached.

Use case 02

Disease resistance pre-screening without greenhouse trials

Tomato Pepper Potato

Inoculation trials for disease resistance are slow, seasonal, expensive, and require live pathogen management. Our Disease Resistance Probability scores are derived from sequence embeddings trained on accessions with documented resistance profiles. Six pathogen resistance scores are returned per accession — Fusarium, late blight, powdery mildew, bacterial wilt, cucumber mosaic virus, and Botrytis.

Pre-screen 1,200 tomato lines for Fusarium and late blight resistance in under 48 hours — without a greenhouse booking, pathogen sourcing, or inoculation protocol.

Use case 03

Yield stability across environments — early prediction

Soybean Sorghum Wheat

Characterizing yield stability across environments requires multi-location trials over 2-3 seasons — significant resource commitment before knowing if a variety is worth advancing. Our embeddings, trained on trial data from multiple climate zones, encode cross-environment stability signals. Varieties whose embedding geometry resembles proven stable varieties are flagged as high-stability candidates.

A stability score before year one of multi-environment testing. Candidates predicted as unstable by embedding geometry can be deprioritised before they consume two seasons of field plot space. Instability is not always visible in single-location year-one data — our model surfaces the signal earlier.

More applications

Additional use cases

Parent selection for crosses

Identify complementary parent pairs by comparing embedding-space distances. Parents whose embeddings are genetically distant but phenotypically predictive of target traits show the highest hybrid vigor potential in our testing.

Crops: Maize, Tomato, Soybean

Diversity panel triage for genebank screening

Genebanks hold thousands of accessions with incomplete phenotype records. Submit sequenced accessions — in VCF or FASTA — to flag which entries are likely carriers of target traits before committing to a greenhouse evaluation cycle that may span two seasons.

Crops: Wheat, Rice, Barley

Marker validation cross-check

Known SNP markers capture partial genetic architecture. Cross-check your current MAS panel against our embedding signal to verify whether a marker set is accounting for the full genotype-to-phenotype relationship or missing polygenic signal before committing to a marker-assisted selection protocol.

Crops: Wheat, Maize, Sorghum

Have a specific breeding challenge?

Let's run a benchmark on your data. Bring a VCF file to a 30-minute technical call — we'll show you what predictions look like for your specific crop and trait targets before any commitment.