HSquared.jl

Appearance

Audience And Comparators ​

HSquared.jl is being built for people who need quantitative-genetic models to work on real breeding and evolutionary data.

Who It Is For ​

  • livestock and aquaculture breeders;

  • plant breeders;

  • evolutionary geneticists;

  • ecological quantitative geneticists;

  • genomic prediction users;

  • applied students and analysts who need readable errors and extractors.

What They Need ​

The package should make common work straightforward:

  • fit pedigree and genomic animal models from R;

  • construct relationship precision matrices without hidden dense conversions;

  • extract breeding values, variance components, heritability, genetic correlations, G matrices, and uncertainty;

  • compare models with diagnostics that distinguish weak evidence from software failure;

  • move from R syntax to a Julia engine through a bridge.

Comparator Discipline ​

HSquared.jl should be tested against the software users already trust: ASReml, BLUPF90, DMU, WOMBAT, sommer, MCMCglmm, JWAS, AGHmatrix, nadiv, and XSim-based simulations where appropriate.

The ambition is to become a free, open, benchmarked engine for pedigree, genomic, livestock, plant-breeding, and evolutionary quantitative genetics. That is a roadmap goal. It is not yet a benchmark claim.

Any public speed or superiority claim must report the machine, software versions, model, data size, number of records, animals, traits, nonzeros, memory use, convergence diagnostics, and the exact comparator estimand.