How it works

The breeding index that fits your farm.

Genemap derives every weight in your breeding index from your operation — your country, your costs, your data. Nothing is averaged across an industry that doesn't match you. Here is exactly how.

Why per-farm matters

An industry index is calibrated against an average producer that doesn't actually exist. A composite breeder running 1,000 cows on hard country with a long-fed grid agreement values a milk EBV very differently from a station selling weaner calves into live export. One coefficient cannot be right for both — and yet, until now, the industry has used one for everyone.

Genemap turns that around. The producer's enterprise structure, costs, market and outcomes are the inputs. The dollar weight on every breeding value is the output, derived freshly for each operation.

The four pillars

Genemap rests on four capabilities that work together. Each on its own is genuinely useful. Together, they produce a per-farm index that no static industry index can match.

Bioeconomic weights, derived from your operation

Closed-form derivation of dollar value per breeding-value unit from your farm profile — breeding system, calving time, joining length, retention, finishing percentage, target grid, gross margin per head.

ii.

Lifetime value of every gene

Each breeding value is valued across its full lifetime expression, with daughter and granddaughter pathways discounted to present value. Maternal traits get the weight they actually deserve in self-replacing systems.

iii.

Closed loop on your slaughter results

Realised carcase outcomes from your processor flow back into the engine. Industry default coefficients give way to numbers learned from your own data. The longer you use Genemap, the more your weights reflect what actually pays on your farm.

iv.

Per-farm genomic prediction — not yet

Not yet built. Genotypes are not stored for this, and no per-farm GWAS or DGV is fitted to your realised profit today.

The maths, in the file.

The four pillars are not slogans. The first three resolve to closed-form expressions in the engine source; the fourth is not yet built. Below are the load-bearing equations, with the code shape that derives them. The full implementations sit in the engine source tree and are open to read.

i. Bioeconomic weight per trait.

The dollar weight applied to each breeding value is the marginal change in gross margin per head if that EBV moves by one unit, multiplied by the lifetime gene-flow expression of the trait.

wi = ∂(GM/head) / ∂EBVi × DGFi

where w_i is the dollar weight on trait i; GM/head is the gross margin per head derived from the producer's farm profile; and DGF_i is the discounted gene-flow expression of trait i.

Reference: Hazel (1943) Genetics 28:476–490 — selection-index theory; Brascamp (1978) Livestock Production Science 5:263 — bioeconomic weight derivation.

ii. Discounted gene flow.

Each breeding value expresses itself over a lifetime — directly in the animal, in its progeny, and in subsequent generations. The platform values the full pathway with NPV discounting.

DGFi = Σt=0..T pi,t × (1 + r)−t

where p_i,t is the probability and proportional expression of trait i in year t of the gene-flow horizon, r is the discount rate (default 5% p.a.), and T is the horizon (default 10 years).

Reference: McClintock & Cunningham (1974) Animal Production 18:237; Hill (1974) Animal Production 18:117 — overlapping generations.

iii. Closed-loop calibration on realised profit.

Industry default coefficients are replaced by numbers learned from the producer's own kill outcomes. A ridge-regularised mixed-model regression of realised slaughter $/head against the animal's EBV vector yields the producer-specific multiplier on every weight.

β̂ = (X⊤X + λI)−1 X⊤y

where X is the EBV design matrix across the producer's slaughtered animals, y is the realised per-head net margin, λ is the ridge regularisation penalty (set by a fixed rule, half the number of kill cohorts in the fit; it is not estimated from the data), and β̂ is the producer-specific multiplier vector that adjusts each industry default weight.

Reference: Henderson (1975) Biometrics 31:423 — mixed-model equations; Hoerl & Kennard (1970) Technometrics 12:55 — ridge regression.

iv. Per-farm genomic prediction — not yet built.

Not yet built. The expression below is the intended form; nothing on the platform fits it today.

DGVi,a = Σj za,j × α̂i,j

where z_a,j is animal a's genotype at SNP j, and α̂_i,j is the per-trait SNP effect estimated by ridge regression on the producer's own animals' realised profit. SNP effects are estimated jointly across traits to retain pleiotropic structure.

Reference: Meuwissen, Hayes & Goddard (2001) Genetics 157:1819 — genomic selection; VanRaden (2008) Journal of Dairy Science 91:4414 — ssGBLUP.

What the code looks like.

A simplified shape of the actual engine derivation. Every coefficient, every fallback, every reason string lives in the source — override any of them on the rank page and the engine re-derives in milliseconds.

// core/js/engine.js — derive bioeconomic weight for trait i
function deriveWeight(trait, profile, country, system) {
  // 1. Marginal effect of one EBV unit on gross margin per head
  const dGM_dEBV = marginalGM(trait, profile, country.grid);

  // 2. Discounted gene-flow expression (years 0..T, rate r)
  const DGF = discountedGeneFlow(trait, profile.system, {
    horizon: 10, rate: 0.05
  });

  // 3. Closed-loop calibration multiplier from producer kill data
  const beta = profile.killCalibration?.[trait.code] ?? 1.0;

  // 4. Production-system modifier (Hanwoo, ungtjur, pelt, dairy ...)
  const sysMod = SYSTEM_MODIFIERS[system]?.[trait.code] ?? 1.0;

  return {
    weight: dGM_dEBV * DGF * beta * sysMod,
    reason: `${trait.code}: ${dGM_dEBV.toFixed(2)} × DGF=${DGF.toFixed(2)} × β=${beta.toFixed(2)} × sys=${sysMod.toFixed(2)}`
  };
}

The sheep maths, equally on the file.

Genemap is a sheep platform as much as a cattle platform. The closed-form derivation framework above carries directly across, but four sheep-specific equations are load-bearing for the sheep audience — the maternal NLW axis, the wool-quality bioeconomic anchor, the maternal-vs-terminal split, and the pelt-primary economic objective. Each ships in the source tree alongside the cattle derivation.

v. NLW — Number of Lambs Weaned per ewe joined.

In maternal sheep systems (NZ pastoral, AU maternal composite, UK lowland), the dominant economic axis is the number of lambs successfully weaned per ewe joined. The marginal value of one additional lamb weaned is the per-lamb sale revenue minus the marginal cost of carrying that lamb.

wNLW = (Plamb × carcassavg) − (costmarginal lamb)

where P_lamb is the per-kg cwt sale price from the producer's processor schedule, carcass_avg is the average lamb carcass weight at sale (~18 kg cwt typical for AU/NZ), and cost_{marginal lamb} is the marginal cost of carrying the lamb from birth to sale (pasture, supervision, vaccination, supplement — typically NZ$22-30/lamb on a hill-country operation, ~A$35-45/lamb in a finishing maternal system).

Reference: Brown, Ball & Banks (2018) Animal Production Science 58: 1421-1431 — Sheep Genetics evaluation methodology; Newman, Dodds & McEwan (2009) NZ Journal of Agricultural Research 52: 359-376 — B+L NZ Genetics methodology.

vi. Wool-quality bioeconomic anchor.

In wool-primary Merino systems, the load-bearing axis is the interaction of clean fleece weight and fibre diameter (micron). The AWEX auction price curve is non-linear in micron — finer wool clears a steep premium — so the closed-form derivation has to be reconstructed differently from any meat-only system.

wwool = CFWa × Pµ(FDa) − Cshear − Ccarry

where CFW_a is animal a's clean fleece weight (kg/year), FD_a is fibre diameter (microns), P_µ(·) is the AWEX micron-to-price curve (a fitted polynomial through the EMI and WMI bands), C_shear is per-fleece shearing cost (A$8-15/ewe), and C_carry is the year-round maintenance feed cost. The derivative ∂w/∂FD is sharply non-zero around the 17-19 micron threshold where the fine-medium premium step occurs. Fleece traits carry no economic weight until AWEX price guides are licensed.

Reference: Greeff, Karlsson & Schlink (2011) Small Ruminant Research 100: 116-125 — wool-quality genetic parameters; Swan, Brown & Banks (2009) Proc Aust Assoc Anim Breed Genet 18: 358-361 — MERINOSELECT calibration.

vii. Maternal-vs-terminal split.

Most AU/NZ commercial sheep operations join a fraction of their ewes to maternal sires (daughters retained) and the remainder to terminal sires (no daughters retained; progeny entirely for slaughter). The selection objective for each sire team is structurally different — terminal sires zero the maternal-trait weights entirely.

wi,total = m × wi,maternal + (1 − m) × wi,terminal

where m is the proportion of ewes joined maternal (typically 0.30-0.45 in AU prime-lamb operations, 0.60-0.80 in NZ pastoral, 1.00 in pure stud), w_i,maternal is the trait weight when the bull is on maternal duty, and w_i,terminal is the same trait's weight on a terminal-sire ram. For maternal traits (NLW, MWT, milk) w_i,terminal = 0; for carcass traits (EMD, FAT) w_i,terminal is amplified to reflect the full slaughter-progeny return.

Reference: Banks & van der Werf (2009) Proc Aust Assoc Anim Breed Genet 18: 326-333 — multi-sire dual-purpose selection theory; Amer, Simm, Keane et al. (2001) Livestock Production Science 67: 223-239 — economic-weight derivation for split-system sheep.

viii. Pelt-primary economic objective.

A small but economically distinct subset of producers (Swedish Gotland, Karakul in Namibia/Central Asia, Icelandic) breed sheep primarily for pelt rather than meat or wool. The lamb is slaughtered at 13-18 kg carcass weight specifically to optimise pelt quality; carcass revenue is residual. The bioeconomic objective inverts.

wpelt = QPK,a × PPK + La × PL + carcassresidual

where Q_PK,a is the animal's Pälskvalitet (pelt quality) score, L_a is the Lockighet (curl/lustre) score, P_PK and P_L are the per-grade premiums from the Swedish pelt auction, and carcass_residual is the residual liveweight carcass value at a 13-18 kg slaughter weight (a much smaller carcass than meat-primary operations would target). The carcass weight is held intentionally low to optimise pelt curl.

Reference: Näsholm & Danell (2007) Small Ruminant Research 71: 110-118 — Gotland pelt genetic parameters; Eythorsdottir & Thorgeirsson (1999) Livestock Production Science 60: 261-267 — Icelandic woolskin quality.

For the worked numbers each of these resolves to on an illustrative example farm in Australia or New Zealand, see the Australian three-systems and NZ pastoral sheep pieces — both walk through producer-fit β̂ multipliers across the relevant ASBV/BV trait sets.

Built in Australia. Calibrated for any country.

The engine is not Australian-specific by accident — it is Australian-built by intent, and globally deployable by design. Add your country and the town nearest to your operation. From there, Genemap reads:

Your local climate — temperature, rainfall, evapotranspiration, derived from your country's national meteorology service where available, with a satellite-and-reanalysis fallback that covers any location on earth.
Your local pasture — forage type, growth pattern, seasonal kg of dry matter per hectare, derived from regional pasture-growth models or remote-sensing greenness indices.
Your local market — fleece traits carry no economic weight until AWEX price guides are licensed. Market prices are MLA's Australian indicators.
Your country's breed evaluations — read and translated automatically. Whichever breed society publishes your animals' figures, Genemap reads them and normalises them into the platform's common vocabulary so cross-country comparisons mean what they should.

Every production system, properly modelled

A grass-finished bullock on the pampas is not the same animal as a young bull finished indoors in northern Europe, or a long-fed Wagyu in Queensland, or a Hanwoo finished over 30 months in Korea. Industry indexes treat them all the same way. Genemap does not.

Each production system carries a different feed-cost model, a different trait-weighting profile, a different grading scheme, and a different economic objective. Genemap supports the major systems globally — pastoral, semi-intensive, fully housed, dairy-beef cross, long-fed Wagyu, Hanwoo intensive, dual-purpose sheep, terminal-sire slaughter lambs, shedding sheep, wool-primary Merino, pelt-primary Gotland and more. The right traits, weighted the right way, for the system you actually run.

The data inside

Genemap consumes every signal available to a modern breeding programme. Some are mainstream. Some are frontier signals that no production breeding tool currently uses. All of them feed into one engine.

DNA — genotypes from Illumina FinalReport or VCF files today; other formats as labs supply samples. Used for pedigree-aware relationships and mate allocation inbreeding control.
RNA expression — transcriptomic markers where available, particularly for performance-trait calibration on adult animals.
Phenotypes — weighing days, structural assessments, reproductive records, slaughter feedback. The data you already collect on your operation.
Climate — daily, gridded, derived from your town's coordinates. Used to model environment-trait interactions and to project feed gaps.
Forage — pasture type and growth pattern, kg DM/ha across the season.
Microbiome — rumen microbiome composition where measured. Maternally-transmitted feed-conversion efficiency proxy where genotyping budget is limited.
Market data — MLA's Australian indicators.
Slaughter outcomes — realised carcase results, per animal, fed back into the calibration loop.

The frontier signals

Genemap carries three signals that no commercial breeding platform currently uses in production. They are schema and modules, not live pipelines. The published literature shows each as load-bearing for forage-system profitability and sustainability.

DNA methylation at known stress-response loci. Heat-shock, glucocorticoid receptor, growth and maternal programming markers. Expressed as an epigenetic stress-resilience score that modifies per-animal economic value.
Maternal fetal programming. Cow nutrition during gestation programs the calf's lifetime productivity. Inputs include dam body-condition trajectory across the trimesters and supplementation events.
Methane intensity as a selectable trait in the bioeconomic objective. Per-animal residual methane intensity, with carbon-credit dollar contribution per progeny when a CO2-equivalent price is available.

Open by design

No proprietary index. No black box. The math is in the file.

Genemap is documented as code, not as marketing. Every multiplier, every reason, every fallback is in the source. Override any coefficient on the rank page and watch the rankings shift in real time. The engine is the product, and the product is fully open to the producer who runs it.