Genemap is built on six decades of published animal-breeding literature, plus contemporary work on epigenetics, microbiome inheritance and methane heritability. Every load-bearing module is documented in source and traced to its citation. Academic teams can read the maths, cite the engine, and collaborate on the next decade of breeding-decision technology.
The platform was built as a working bioeconomic engine for producers, but the architecture deliberately surfaces the kind of structure academic researchers care about: a fully open methodology, an AI translator over the global long tail of breed evaluations, and a calibrated learning loop that runs in production every 24 hours.
Every coefficient ships in the source tree. Selection-index theory (Hazel 1943), discounted gene flow (Brascamp 1978), BLUP (Henderson 1975), ssGBLUP (Meuwissen Hayes Goddard 2001, VanRaden 2008) — all implemented, all cited, all overridable on the live rank page.
An LLM-backed Tier 2 translator normalises any breed-society's evaluation output into a common semantic schema — EBV / EPD / DEP / ASBV / Zuchtwert / DEP / BLUP-EPD all resolve to the same trait vocabulary. Useful for any study comparing across countries or evaluation systems.
Ridge-regularised regression of realised slaughter $/head against EBV vectors per producer. Per-trait DGV fit against profit. The full calibration loop runs nightly across the network and the math is reproducible from the source.
Pastoral, intensive housed (ungtjur), Hanwoo, Wagyu long-fed, pelt-primary Gotland, dairy sheep, shedding sheep, wool-primary Merino and more — each with its own trait-weight profile. Cross-system comparative work uses these as the explicit parametric overlay.
Three frontier signals are first-class evaluation axes — methylation at known stress-response loci, rumen microbiome composition, and per-animal residual methane intensity priced against country-specific CO₂e markets.
Native ingestors for 18 countries' breed evaluations, climate sources and market feeds. AI translator covers any other country. Useful for cross-country studies that previously required individual MoUs with each breed society.
We make a distinct data tier available to peer-reviewed academic groups under a research-use agreement. The terms are designed to support reproducible publication while protecting individual producers' commercial data.
Anonymised per-country, per-system, per-trait anchors recomputed nightly across the producer network. Minimum 5-producer bucket size. Suitable for cross-country comparative work and longitudinal calibration studies. Available on request.
Plausible synthetic producer profiles drawn from the realised distribution, with full EBV vectors and realised-profit outcomes. Designed for methodology validation without exposing real producer data. Available on request.
Paired native + AI-translated EBV records across 18 country breed evaluations, with semantic-mapping confidence scores. Useful for benchmarking other LLM-translation approaches in animal genetics. Open licence on request.
The closed-form bioeconomic derivation, the ridge-regression calibration code, the production-system modifier tables — all readable in the source tree. Pre-publication access for review by collaborating teams. Available under collaboration MoU.
Until a peer-reviewed engine paper is published (expected H2 2026), please cite the methodology page directly. The citation block below covers the canonical engine derivation; cite individual underlying papers (Hazel, Henderson, etc.) separately as appropriate.
The full bibliography of work the engine builds on is at research.html — 34 references across selection-index theory, BLUP, ssGBLUP, mate allocation, epigenetics, microbiome, methane and pelt-primary genetics. Every citation maps to the engine module that implements it.
The roadmap calls out areas where the platform infrastructure is ready but the published validation work isn't. We're particularly interested in academic partnerships on:
If your research touches any of these and you'd like to discuss collaboration, reach out via research@genemap.com.au.
Genemap's country-aware vocabulary system normalises 15 canonical livestock terms (stud, mob, paddock, abattoir, shedding sheep, EBV, ute, etc.) into per-country preferred forms. Pick a country to watch all 15 resolve live — useful for understanding how the platform handles cross-country terminology in its UI.
Source: core/js/vernacular-resolver.js (VR.1). DOM spans marked <span data-vt="stud"> auto-localise on page load when MGT_USER_COUNTRY is set. See the architecture page for how this fits into the layered engine.
The platform is built to be read, not just used. Open methodology, open citations, open data on request, and an active research-collaboration inbound for academic teams.
Read the methodology See the bibliography